The markers also overlap with other neurological diseases and can be affected by other clinical conditions and the medications used to treat them.
Says Dr. Shaw: “One of the things that people in this field realize is that there are very few Alzheimer’s patients with quote, unquote ‘pure’ AD, with no other pathologies.” Roughly 30 to 40 percent have concomitant Lewy body pathology, he notes, while some 20 to 30 percent have a TDP-43 copathology, such as amyotrophic lateral sclerosis, frontotemporal dementia, or both. Neurovascular disease is also fairly common.
“In other words,” Dr. Shaw says, “our brains are never really that pure in any respect.”
Other complicating examples, says Dr. Frykman, include chronic kidney disease, which, due to reduced kidney clearance, can lead to increased protein levels. While it may not have an important clinical implication, “it still has to be sorted out.” In addition, blood biomarker levels can be affected by obesity or use of GLP-1 analogues. Some blood pressure medications can also affect levels, though this is rare. The first-generation tests also appear somewhat susceptible to ALS, resulting in high peripheral p-tau levels unrelated to brain. And tau can also become elevated following a concussion.
Given the overlapping scenarios, says Dr. Tran, “we’re going to have to figure out how they play out and how they perform for each of these diseases. And as we start looking at a combination of these biomarkers, from a blood or CSF testing standpoint, how will they interact with other technologies, such as digital biomarkers,” which use software to identify potential cognitive impairment and may point to the need for follow-up blood testing.
Dr. Tran hopes AI might be useful to help manage the onslaught of information. “Obviously I’m biased,” he says, “since I work for an AI center.” (He is senior technical director for AI technologies and operations, Computational Pathology and AI Center of Excellence, University of Pittsburgh.) But aggregating multiple, complex factors—including, possibly, information about patient lifestyle and socioeconomic status, as well as environmental and geographic data, etc.—might provide useful context for using current and future biomarkers.
As these biomarkers wend their way into primary care practice, laboratories could and should become the bridge to nonspecialists. Particularly in the early days—à la biomarker spring training—the overriding issue is physician education, says Dr. Tran.
“That’s a big starting point, especially since 181 p-tau will be available to primary care physicians.” Again he suggests AI could be valuable, alerting clinicians that a result could be falsely low or high depending on the patient’s comorbidities—it can be elevated in patients with Down syndrome, for example. (Of course, Dr. Tran notes, those with Down syndrome can also develop Alzheimer’s.)
The main question Dr. Frykman reports hearing from clinicians is “whether the test is good enough.” This is the C-major chord of almost any test—an obvious but basic starting point. They also want to know about slight elevations above the positive threshold, he says. Since the proteins are very low abundance, it probably still reflects a true value. But the presentation and family history are critical elements to consider as well and need to be assessed by experienced clinicians. They also have questions about very high results: What do they mean? Are the results wrong? If a positive ratio test yields a discordant (i.e. negative or intermediate) result in p-tau217, is one element of the test more trustworthy?
“Clinicians know about the test, but they’re not specialists in terms of what can go wrong, how to think about the test,” Dr. Frykman says. “So they take the numbers at face value, which is understandable—that’s not their line of work. So we have to be quite active in leading them down the right path.”