Precision neurology: Why Alzheimer’s diagnosis requires more than one biomarker

• Blood-based biomarkers offer a highly accessible signal for identifying Alzheimer's pathology, a critical input within a holistic differential diagnosis 
• Moving toward true precision medicine requires identifying the specific underlying brain pathologies driving a patient's symptoms instead of relying on broad clinical categories 
• Comprehensive testing panels with rapid laboratory turnaround times significantly mitigate patient anxiety by reducing diagnostic uncertainty

The recent advent of blood-based biomarkers for Alzheimer's pathology promises to offer an accessible biological signal that can be measured with a simple blood draw. In this exclusive LabLeaders interview, Craig Ritchie, Professor of Psychiatry of Ageing at the University of Edinburgh, reflects on this new technology. He argues that a precision-neurology pathway should combine patient and family history, cognitive assessment, evaluation of comorbidities and co-pathologies, imaging, and appropriately selected biomarkers. The implication is not simply to add a test: It is to define where testing enters the pathway, how reliably and quickly results are returned, what clinical context accompanies them, and what happens after a positive, negative, or indeterminate result.

Cognitive impairment has many different possible causes, so the first step is a detailed history from the patient and family—not an automatic assumption of Alzheimer’s disease. When people present with cognitive impairment, whether to primary or secondary care, the most important thing is speaking to the person and speaking to their family about the trajectory of that decline over time. Was it a sudden onset? Has it been building up for years? What other impacts has it had? It's a mistake to always assume that cognitive impairment, particularly in an elderly person, must be due to Alzheimer's disease. What we have to recognize is that it very well may not be due to that, and taking a good history is the first and most important step.

There's so much awareness now in the public of Alzheimer's disease as a memory disorder that people's first fear will be that it's due to Alzheimer's—and that can overshadow a lot of other considerations about what might be causing it. One of the other big challenges is that cognitive function declines anyway as we get older. So there are two sides of the same coin: Some people present early, fearful and worried about their cognitive health, their brain health, and other people present very late because they think this is just a normal part of aging. Both are worth considering.

A blood biomarker should change the probability of Alzheimer’s pathology within a wider diagnostic workup; it should not be treated as a standalone diagnosis. One has to realize that Alzheimer's disease, or any other neurodegenerative disease, is actually a very complex disorder. In some ways, we've found ourselves in a position where we think the diagnosis is easy, because there's such demand on services that we want to do all of our assessments and make our diagnosis as quickly as possible. I always reflect on other conditions like cancer: You wouldn't make a diagnosis of cancer on a single test in half an hour, but that's sometimes what we do in a memory clinic. Relying on a single test is simply bad medicine. When anybody presents with any condition in medicine, you always have a differential diagnosis, and what you do with your history-taking, your physical examination, a review of other medications, and proper cognitive testing is to change that differential—some things become more likely, others less likely. Part of that is doing a blood test for Alzheimer's pathology, because that helps move your probability up and down depending on the result. But it's not done in isolation from all the other things.

A holistic workup must assess both comorbidities that can affect cognition and co-pathologies that may be driving neurodegeneration. At the moment, people tend to present once they already have cognitive impairment. They're not presenting, if you like, as a high-risk population concerned that they might develop memory problems. I'd like to think there will be a point in time where people come in before symptoms emerge. But right now, people present because they feel their own cognition falling away, or because those around them have observed cognitive changes. The workup is to recognize that there are both what we call comorbidities and co-pathologies. The comorbidities are other things that could be driving it: Depression, stress, and poor sleep. We want at least an initial assessment of those factors. Then you have the co-pathologies: You want to see whether the person is getting a bit of parkinsonism, for instance, or has a trajectory of change in their behavior, cognition, or language that might suggest more of a frontotemporal picture. So that assessment, as far as possible, should include all of the comorbidities known to influence cognitive health, as well as the co-pathologies that might be driving it.

The ideal assessment would follow a series of triage points. You're not going to do everything for everybody right at the beginning. A general practitioner (GP) isn't going to organize a lumbar puncture or a PET scan, so you need to recognize where in that journey the person is, and where the likelihood of a diagnosis is highest. In primary care, everything is on the table; by the time you reach secondary care, you should already have removed certain differential diagnoses. But what's often missing, even in that setting, certainly in the UK, is biological evidence of what might actually be causing the disease. In the UK, for instance, fewer than half of the people in a specialist center get a brain scan, which to my mind seems incredible: That you could be assessing what is a very serious brain disease without doing any brain imaging at all.¹ There's been a historical reliance on neuropsychological testing, examination, and history-taking without having a good look at what's actually happening to the brain. We often find that standard tests, for things like thyroid disease or B12 deficiency, are done because they're easily done, and can be done by the GP. But other things, like brain imaging or amyloid testing, are often not done because they're expensive, because people don't see the value of them, or because they're challenging, like a lumbar puncture. And I think we're only partially sighted if we don't have a good look at what's actually happening inside the brain.

I go back to what I was saying about comorbidities and co-pathologies. The comorbidities are easy: High blood pressure, diabetes, nutrition, and so on. The co-pathologies are harder, because once we get to the point where we believe this is a neurodegenerative disease, the differential could include Lewy body disease, vascular, cerebrovascular disease, Alzheimer's pathology, frontotemporal pathology, etc. There are clinical features of these conditions that may lead you to think it's one over another, but in an ideal world, we'd have a specific biomarker for each of them. We've started with Alzheimer's pathology, but alpha-synuclein, TDP-43, MTBR—these are the things I hope will be in clinicians' repertoire in the near future, so we can say what the brain pathology actually is that's driving those symptoms. Cerebrovascular disease is a huge aspect of this, and fluid and imaging biomarkers are in development that will be much better at defining this disease’s contribution to a person's presentation.

At the moment, we tend to say it's Alzheimer's dementia, it's mixed dementia, or it's vascular dementia, which to me is intellectually redundant. Putting people into those huge categories on the basis of clinical phenotype is, to put it best, inaccurate. If we're going to use medicines to treat people, they need to be quite specific to that pathology and move toward more precision medicine. You can't have precision medicine for clinical syndromes. And the management is different for each of these pathologies; the management in the future will be very different. There'll be medicines like in cancer, where it's not monotherapy for most cancers—you'll have very judicious use of targeted therapies depending on the pathology that's driving that person's symptoms. We need to make sure people know these are different diseases: They might have similar symptoms, but you don't just identify a single category and say it's all the same. The other thing about precision medicine is that it also gives you a better idea of the prognosis, because the trajectories of change aren't identical between different conditions, nor are they identical for different people who have the same condition. So there's work to be done on better prognosis too. Fluid biomarkers, genetics, and other comorbidities are probably going to help with that.

We've lived through, and are still in, a period of people being on waiting lists for a very long time. There's a lot of anxiety when patients are given a diagnosis of mild cognitive impairment, because we often see that people given this diagnosis kind of knew it already. They knew they had a mild cognitive impairment, and then they ask: "What does this mean? What's going to happen to me?" The usual answer is, "Well, we don't know, but why don't you come back in a year and we'll see what's happened?" So even after the diagnosis, that lack of certainty about what is driving the illness, and what is likely to happen, increases anxiety quite substantially. You can be on a waiting list, anxious and worried; and even after you've been seen, the diagnosis sometimes doesn't help you understand what the next stage of the illness is going to be. Earlier use of tests that help clarify the etiology helps reduce that anxiety. It might not be good news, but that's true for cancer as well. A lot of people worry about their symptoms and what's going to happen to them, but anxiety is usually born out of uncertainty: People are anxious because they're uncertain about what's going to happen. If you can tell them what you think is most likely, with a fair degree of accuracy, then that uncertainty goes, and the anxiety reduces. Not giving an answer from a test doesn't make the pathology go away. It's better to know than not to know.

A result is useful only when it is analytically reliable, returned in time to inform the next decision, and connected to a clear clinical pathway. We're in a new world at the moment, where we haven't really started rolling out these tests, like blood-based biomarkers, at scale, at a population level. There are obviously innovative centers who are using them more than others. But when you think about waiting lists and about waiting for test results to come back, the length of time between the test and the result should be as short as possible. There's no advantage to waiting. If you think about decision-making in the emergency department, on troponin levels and point-of-care testing, people don't want to wait around for a test result, particularly when the consequences of that result are so meaningful to them. So automation and rapid throughput matter, and you get more throughput when more people are asking for the test, so you're optimizing your platform and your analyzers. I think the turnaround time should eventually be measured in days, rather than the weeks and months it can be at present.

It's such an important question because we shouldn't look at a test in isolation from other factors and features. Some of what I've already said, about comorbidities, sleep, anxiety, depression, these things are actually very easily assessed. You can use rating scales, you can use digital platforms. You don't have to spend hours in an interview assessing sleep. You can give somebody a Pittsburgh sleep questionnaire. You can do a lot of these things online, with the patient in the waiting room. So there's no real impediment to having a comprehensive assessment in a timely way. And, coming back to what we said earlier, if you can have a panel of biomarkers where you understand their performance, their sensitivity and specificity, and so on, then this shouldn't be a long, drawn-out assessment. It's not surgery, doing a tissue biopsy; it's a blood test and some online testing.

Laboratories can add value by interpreting results in clinical context and maintaining a two-way dialogue with care teams, not just returning the numbers. I'm often struck by conversations I had 10 or 15 years ago with my colleagues in radiology. They would often get an MRI request that just said "query dementia" — so when they were looking at the brain scan, that was the only information they had. With a neuroradiologist, as it should be with a laboratory, it's a dialogue. The more information you give to the people who are assessing the result in the lab, the clinical biochemists, or whoever it might be, about the patient's cognitive status, their history, and so on, the better. It's more than just giving a value back, it's contextualizing that value amongst everything else. Now, I'm not saying you send a 20-page report to the lab, but there are probably some features in that context that would help make the laboratory's clinical biochemist a little more accurate in interpreting the result. That clinical biochemist isn't the person sitting in front of the patient. The information still has to be communicated by the practitioner to their patient, but I think about things like biochemistry and hematology: If there's an array of results from the hematology on the MCV, the MCH, the hemoglobin, the platelet levels, you sometimes get a hematologist phoning you afterwards. We need the same type of communication when it comes to neurodiagnostics as well.

I think the last thing to say is that I see a future where we can monitor Alzheimer’s much like we do with cholesterol. If we can see the amyloid levels, we can help patients to prevent Alzheimer’s. We can then recommend lifestyle changes and appropriate treatments. If we can prevent Alzheimer’s through very early detection with a blood test, we have found something functionally better and more realistic than a cure.