This story originally ran on Dec. 1.
Name: Reisa Sperling
Position: Associate Professor of Neurology at Harvard Medical School Department of Neurology, Brigham and Women's Hospital; Chair of the National Institute on Aging/Alzheimer’s Association Workgroup on Preclinical AD.
Background: MD, Harvard Medical School; Assistant Professor of Neurology, Harvard Medical School
Reisa Sperling is an associate professor of Neurology at Harvard Medical School where she focuses on research into Alzheimer’s disease. With Mayo Clinic researcher Clifford Jack and University of California, San Diego, researcher Paul Aisen, she published this week a commentary in Science Translational Medicine suggesting that the recent failures of several anti-amyloid Alzheimer’s drugs in clinical trials stemmed from starting the therapies too late in the progression of the disease.
Among the topics the article discussed was the potential role for protein biomarkers in drug development, particularly in putting together appropriate clinical trial cohorts and monitoring therapies’ effectiveness.
This week ProteoMonitor spoke to Sperling about the current use of protein biomarkers in Alzheimer’s drug research and where the field might be heading.
Below is an edited version of the interview.
In your commentary you outline a number of issues that have stymied the development of drugs for Alzheimer’s disease. Where in the process, specifically, do you think protein biomarkers have the most potential to help?
I think there are likely multiple problems [facing Alzheimer’s drug development], but the problems we were focusing on were that we were trying a specific class of drugs, anti-amyloid agents, at a relatively late stage of the disease where amyloid is likely less relevant.
So there are two challenges where I think biomarkers and protein biomarkers specifically can play a role: One is trying to identify individuals who are in the early stages of the Alzheimer’s disease process, likely before they have symptoms at all – so for that we need biomarkers to identify them. The second place where I think protein biomarkers can play a role is in tracking decline. One of the difficulties in normal individuals – whether they’re genetically at risk or amyloid positive – is how do we tell if our drugs are working? Are they changing the downstream neurodegeneration? And since they don’t have symptoms yet, and it may take a long time to track changes in clinical disease, we need markers. So I think that protein markers that tell us about the amyloid-related process and downstream neurodegeneration are going to be needed.
How are such patients currently monitored during drug trials in terms of determining a treatment’s effectiveness?
Trials for Alzheimer’s disease and dementia are monitored primarily with clinical outcomes. In the past year or two, you have definitely seen some studies that are incorporating biomarkers – spinal fluid, amyloid imaging, structural MRI. There are sometimes plasma proteins that are used as exploratory measures, but they’re not typically used as primary outcomes.
There are trials in what we call prodromal Alzheimer’s disease, or mild cognitive impairment, and those trials are incorporating biomarkers even more robustly. They are using biomarkers both to help identify patients to come into the trials and to track response, because, again, the earlier you look in the disease, the more you need these biomarkers to help you.
You noted in your commentary that for anti-amyloid therapies, even the prodromal stage might be too late to start treatment.
Right, because I think that at this prodromal or mild cognitive impairment stage – although it’s probably better than AD dementia – there is increasing evidence that by that stage of the disease there is already significant neuronal loss. Even in very early MCI you can already see neuronal loss. I just think all of these data are suggesting that the disease is well entrenched even by the prodromal stage, so I think we have to go before there are symptoms.
I think it’s very clear that in CSF, biomarkers at the stage of MCI are already very clearly abnormal, and, in fact, most data suggest they are turning abnormal before people have MCI.
Much of the Alzheimer’s protein biomarker research has revolved around A-beta, tau, and phospho-tau. Why have those markers emerged as a focal point?
A-beta is important because all of these therapies are targeting amyloid, but the downstream markers, both tau and phospho-tau, are markers of neurodegeneration, and although we think that is somehow related to amyloid, it’s not really known how amyloid causes or is related to neurodegeneration. But we do know that tau and phospho-tau in particular are abnormal, certainly, by the stage of mild cognitive impairment. And there has been a series of papers over the last year suggesting that amyloid-positive normals also are likely to have elevated tau, suggesting that they’re on the road toward Alzheimer’s.
There’s also been work on a number of other Alzheimer’s markers – auto-antibody-based approaches and plasma protein panels, for instance. How much traction have protein biomarkers outside A-beta and tau achieved in the industry?
There’s been a tremendous amount of work. A lot of those are very promising. In terms of traction, I think the issue is being able to standardize them for use across labs. Even A-beta, tau, and phospho-tau have been notoriously difficult to standardize across laboratories and even within laboratories across runs. So I think that some of these other biomarkers are very promising in one person’s laboratory. The hurdle now is to look at their reliability and reproducibility – these kinds of things. So I think they have traction, but I suspect it’s going to be a while before we see any of these other ones in widespread use given that people are still fighting over how to standardize even the now commonplace markers of A-beta and tau.
How are the efforts to standardize A-beta and tau coming along?
It’s progressing. There’s been actually quite a bit of work. There’s definitely been progress. And I credit a number of international groups, especially [University of Gothenburg researcher] Kaj Blennow, who really has worked to gather samples and send them around to multiple labs. There has been a series of publications over the past year that suggests that the reproducibility is getting better.
But it’s still problematic, and in particular the issue is being raised that you need some kind of standard to include in every run, and that’s really a problem in clinical trials where you can’t wait until you get all the samples and then run them. You have to be able to run them in real time for recruiting people into studies. Potentially with outcome markers you might be able to hold them all and do them all at once, but generally you have to be able to do them in real time, and if there’s significant variability across the assay run to run, that’s really a problem. It’s getting better, but it’s not perfect yet.
Most of this work is going on in CSF. What are the difficulties involved with pursuing blood-based biomarkers for Alzheimer’s?
I think there are some interesting plasma proteins, but I do think its partly right that with the critical proteins that are markers of what is happening in the central nervous system [there is the question of] how much of them are passing over the blood-brain barrier into the blood, and then I think the question is when they do cross are they rapidly getting metabolized or degenerated or tagged with antibodies or other things that are taking them out of the system? So I think it’s hard to say. Are they not crossing so you don’t see them, or are they at such low levels, or are they’re just not there for long?
I don’t think we should give up on plasma markers, because there has been some work, at least on A-beta, suggesting that longitudinal changes within someone may be a marker of future decline. There have been a couple of studies that suggest that if you have changes in your plasma A-beta over time and you’re still normal, that, unfortunately, predicts that you’re more likely to decline clinically five to 10 years later. The problem is whether you can set some threshold level between individuals and use it as a diagnostic measure or an outcome marker that way.
How widely have pharma firms incorporated protein biomarkers in their Alzheimer’s drug development efforts?
Multiple companies are now doing this. They are widely used at the stage of MCI. As trials are moving from the dementia stage to the prodromal or MCI stage, these biomarkers, mostly in spinal fluid, are being incorporated much more typically. So most of the MCI trials with anti-amyloid agents right now are requiring evidence of amyloid positivity in either spinal fluid or amyloid imaging to come into the trial. And as we move even earlier to secondary prevention trials in normals or genetic at-risk patients, [biomarkers] will absolutely be required. So this trial idea that we put in the commentary of using amyloid-positive normals as a secondary prevention group, that group would be purely defined on the presence of protein biomarkers.
Because that’s the only way you could define them given that they are normals?
Exactly. So [the trials] will either use [measurements of protein biomarkers in] spinal fluid or amyloid imaging. Anyone who is really contemplating an MCI trial right now is incorporating biomarkers. In that case it’s not just to find the subjects as it would be in [a trial studying] normals; it’s to make sure that the cognitive impairment is due to Alzheimer’s disease and not to something else.
When did the use of protein biomarkers in these sorts of drug trials become standard?
In the past year. It’s not been a long time. Most of these prodromal trials have started in the last year or two, but I think there were a series of MCI trials that were begun ten years ago, many of which were negative. As the field has moved towards having more specific targeted therapies toward amyloid, it’s become clear that you need to make sure that the patients you are recruiting into the study have the target pathology. Because, again, if you’re using an anti-amyloid drug, you need to be sure that the cause of the MCI is Alzheimer’s disease, and that they have the target, which is amyloid.
What role do tau and phospho-tau have to play?
I think that tau and phospho-tau have a huge role to play as outcome measures and potentially as inclusion markers as well. The vast majority of MCIs who have amyloid have tau. In normals sometimes you see amyloid without tau, but in MCI, if they have amyloid they have tau almost always. We are starting to get tau drugs coming into the clinic, but the vast majority of studies that are ongoing, at least at phase II or phase III, are anti-amyloid, so the critical thing for including them in the trial is making sure they have amyloid pathology. Tau and phospho-tau would be critical, though, for saying, 'Are you changing the course of the disease?'
Are there any regulatory or ethical issues with regard to trials on amyloid-positive normals? Can you enroll otherwise healthy patients in a drug trial based on just a protein biomarker?
I think that there is already evidence that on a population basis, having amyloid increases your risk of having cognitive decline. What isn’t clear yet is whether at an individual subject level we have enough information to say if a person will decline and over what timeframe. However, what I hope we argued successfully in the article is that part of the way we will get this answer is by doing these trials.
One way [to approach this question] is you identify thousands of people who are amyloid-positive and normal and wait to see who declines and what does that mean. That’s going to take a decade. We absolutely have to do those studies, and I’m doing one of them – the Harvard Aging Brain Study is exactly that, looking at people with and without amyloid and asking if there are differences in rates of decline. But another key piece of evidence is what happens if you remove amyloid, if you lower amyloid — does that change the likelihood of decline? And I think we have to start those studies as well. We can’t wait another 10 years to start those drug trials. It’s just too much of a critical problem to wait.
In your commentary you analogized it to cholesterol’s role in the cardiovascular space.
Exactly. Most people with high cholesterol will never have a heart attack or stroke. But we still lower cholesterol on everybody because maybe we prevent 1 in 100. On a population level that makes a huge difference.
We’re clearly not there in Alzheimer’s disease yet. We clearly don’t know the risks of amyloid and cognitive decline. But I think its important to look at the cardiology world, because if you look back they had what they called the “cholesterol wars” about 30 or 40 years ago, and people are now bemoaning it, saying that changing cholesterol only accounts for a 20 percent reduction in heart attacks. Well, imagine if we could reduce Alzheimer’s disease by 20 percent. I’d be thrilled. Amyloid is not the only answer, clearly. It’s probably one piece in a very complicated puzzle. But if we could modulate that piece, and we could do it early enough, maybe that’s enough to at least make a dent.
Over the last year, groups like the International Working Group for New Research Criteria for the Diagnosis of Alzheimer’s Disease and the National Institute on Aging have published new guidelines that make more use of protein biomarkers in defining and diagnosing Alzheimer’s disease. Is this a sign the field is moving in this direction?
I believe that at least in research, it’s clearly where the field is moving. The question is, how do we get to the point where we can translate what we’re doing in research into the clinic? And we’re not quite there yet. So the preclinical criteria are really for research only right now. At the stage of MCI, it’s a little bit of a grey zone because the evidence is continuing to mount that levels of these protein biomarkers in spinal fluid are predictive of who will decline. I think that the way it’s laid out in these working group documents where if you are positive for both biomarkers – both amyloid and neurodegeneration, so either tau or MR atrophy or any of the downstream markers – the likelihood that this is Alzheimer’s disease is very high. But I think we probably need more research before we can say this is what should happen in clinical practice.
Have topics you'd like to see covered in ProteoMonitor? Contact the editor at abonislawski [at] genomeweb [.] com.