Mark Herz: Good morning. This is GBH’s Morning Edition. Millions of Americans either have or are touched by Alzheimer’s, and there’s great hunger for treatment, for prevention, for hopefully and eventually a cure. Well, we have a piece of local hopefulness to talk about today. With us to tell us about some new research is Noureddine Melikechi, professor of physics and dean of the Kennedy College of Sciences at UMass Lowell. Welcome, professor.

Noureddine Melikechi: Thank you. Thank you so much for inviting me.

Herz: So this is your research, and it’s a really kind of complicated and fascinating-sounding series of steps that you and your team took to discover what are called "biomarkers," or they’re potential biomarkers. Can you break that down for us laypeople?

Melikechi: So, as you may or may not know, Alzheimer’s is diagnosed through a platform of biomarkers. And all of these proteins are — a diagnosis is done through very complicated, highly experimental protected PET imaging. And through CSF, through taking the cerebral fluid. And that makes it hard and not available to everybody.

The other piece to this is there’s actually some evidence from studies that have been done in the past that some metals and some atoms and some ions that we have in our body — sodium, potassium, magnesium, copper and others — do play a role. And the studies that people have done show that almost individual metals have a role to play. So they play a role in Alzheimer’s.

What we show in our study is that actually the individual metals — such as sodium, potassium, magnesium, again, copper, and all of these — although they play a role, what’s more important is the relationship that they have among each other.

So the study shows that we can look at some biomarkers among all of these ions that are in our blood, right, and look for these relationships. And when you find these relationships, then actually you can find that you can classify blood, not CSF, from patients that are healthy, and those who have Alzheimer’s.

So what does this do? It does two things. One, it gives you a new way of trying to understand the disease. So, again, it’s not just levels of these ions, but the inter-relationship between them. And second, it can be done through blood, rather than doing it through CSF and it can be done without having million-dollar equipment to be done. So trying to do something really simple that will be accessible to everybody, that’s the goal here.

Herz: It sounds like it’s an exciting example of how AI has the potential to advance medicine and in this case, medical research specifically. So tell us about how you used AI and how you see it contributing to what you do in particular.

Melikechi: The hypothesis is that with AI coming into here, it gives us — it’s a new tool that allows us to look at many, many variables at the same time. So, you know, the human brain maybe can take seven variables or something like that and try to understand them. But with the power of computers that we have with the software that’s being developed in AI, AI allows us to look into this vast array of variables — and the body is very complex, really, really complex — and try to find the relationships.

So what we’re trying to do is: [find] relationships between these atoms, at the fundamental level, at the atomic level; and the proteins that exist in our blood; and the disease. So it’s a path of going from the basic fundamental structures that we have, you know, that make up all of us, and the proteins that these molecules, these atoms help make, and the diseases. So instead of looking at one alone, this study, what we hope, is to find these paths to diseases. And by finding those paths of diseases, then hopefully we could understand them better. And we can then cure them and so forth.

So AI gives us that ability to look at so much data. But I have to say one more thing here: AI, for us, and as a physicist, we have to be careful that it’s not just a black box. So when we do all of these things, when ... AI gives us the results, then we go back. Typically in my group, we go back to the original data that we have. And instead of looking at the thousands or 2,000s, we look exactly at what are those small, those limited number of variables, and try to find the whether the conclusion is valid or not or is valid to a certain point. So that’s what we do.

Herz: Why is a physicist doing research on a medical disease?

Melikechi: And why not?

Herz: OK.

Melikechi: So the approach we have here is are trying to see, again, because you and I and these things — the table, the chair I’m sitting on, it’s all made up from atoms. So, at the beginning, it’s atoms that starts that. And those atoms become molecules that become, you know, become proteins, become cells, etc., etc.. So, why ignore the fundamental aspect of things?

So our research essentially, I would like to say, is like having, you know, you need all of these things to work. You know, the protein levels, the atomic levels. It’s like music. So if you have them all working very well, it works. But if one of them goes down or goes up, then the hypothesis is, then it triggers a disease: It could be cancer, it could be Alzheimer’s, it could be something else.

Herz: Well, do you have an idea of how much promise there is in these initial discovery of these biomarkers? Like Where it might lead in terms of treatments?

Melikechi: Sure. So as scientists, we need to know what our conclusions are. We also need to know what the limitations of the conclusions are. And the limitation here is that we had just 100 samples, 100 patients that we worked on, roughly speaking. This one — we did another study as well, which gave the same results. So really, it’s not just about the number of patients we need to look at, but to make sure that actually the uncertainties that we had on on the result, we put a cap on it. So that’s why you see in what we publish, we said this is true too, up to this level. So I think there’s a lot more work that needs to happen, needs to be done, before one can say we can develop a device and so forth. How long would it take? It depends on funding. It depends on how much effort we put into it.

Herz: And you think this could lead to a cure and not just a treatment to sort of, you know, ameliorate things? Why do you think that is?

Melikechi: I think every time you understand a disease in this case, right, or process at its fundamental level, at what actually triggers it, not just, you know, seeing it when you have symptoms, but before, before it gets there. I think it provides new tools for other groups who may then use those tools to develop new treatment, new drugs, new whatever that might be. Again, that might take a while, but I think by letting people know that, for example, relationships between copper and iron and selenium and sodium and potassium are important — maybe there’s this music. Somebody cleverer than me can make something that can tune us in such a way that we are doing something that would help us.

Herz: Here’s hoping. Yeah, yeah. Really interesting work you’re doing. Noureddine Melikechi, professor of physics and dean of the Kennedy College of Sciences at UMass Lowell. Thank you so much for joining us.

Melikechi: Thank you.

Herz: This is GBH.

Research from a UMass-Lowell team found that Alzheimer’s patients had certain concentrations of common elements in their blood plasma not present people who did not have dementia — giving researchers potential biomarkers to look for as they work to better diagnose and find treatments for the disease.

The research offers a look at what Alzheimer’s looks like at the atomic level, said Professor Noureddine Melikechi, dean of the Kennedy College of Sciences at UMass-Lowell and an author of the research.

“Those atoms become molecules, that become proteins, become cells, etc., etc.,” Melikechi told GBH’s Morning Edition. “You need all of these things to work: The protein levels, the atomic levels. It’s like music. So if you have them all working very well, it works. But if one of them goes down or goes up, then the hypothesis is, then it triggers a disease: It could be cancer, it could be Alzheimer’s, it could be something else.”

About 7 million Americans live with Alzheimer’s disease, according to the Alzheimer’s Association. While there are some FDA-approved medications that can slow down the disease’s progression, there are no known cures.

Related Stories

Melikechi and his colleagues took blood samples from 25 Alzheimer’s disease patients at a Bedford dementia care unit and 34 people who do not have signs of dementia. They then analyzed their blood plasma looking for concentrations of certain elements: Sodium, iron, copper, phosphorus, magnesium, zinc and potassium.

“There’s actually some evidence from studies that have been done in the past that some metals and some atoms and some ions that we have in the body — sodium, potassium, magnesium, copper and others — do play a role [in dementia],” Melikechi said.

The exact role they play — and how people can harness those processes to prevent or treat the disease — is yet to be determined conclusively, Melikechi said. Scientists will need to conduct a lot more research into these potential biomarkers before dementia patients and their families can expect to see changes in how doctors prevent or treat Alzheimer’s.

But this study offers potential advancement in two areas, Melikechi said: It gives researchers another way to understand the disease, through the ways these elements interact. And it opens a new avenue for diagnosis — blood tests, as opposed to current methods, which sometimes rely on more expensive, complicated, and hard-to-get tests like examinations of cerebrospinal fluid.

“It can be done without having million-dollar equipment,” Melikechi said. “Trying to do something really simple that will be accessible to everybody: That’s the goal here.”

To analyze their samples, Melikechi and his team used machine-learning algorithms. That allowed them to look at the interplay between a larger number of variables, he said.

“What we’re trying to do is relationships between these atoms, at the fundamental level, at the atomic level, and the proteins that exist in our blood — and the disease,” he said. “The human brain maybe can take seven variables or something like that and try to understand them. But with the power of computers that we have with the software that’s being developed in AI, AI allows us to look into this vast array of variables — and the body is very complex, really, really complex — and try to find the relationships.”

While AI was a useful tool to his team, Melikechi said they were careful to double-check its work by looking more closely at the variables that it flagged.

“We have to be careful that it’s not just a black box,” he said.

Melikechi is a physicist by training, not a medical doctor, so his approach to looking at diseases in the human body stems from looking at the basic building blocks of all matter.

“I think it provides new tools for other groups who may then use those tools to develop new treatment, new drugs, new whatever that might be,” he said. “That might take a while, but I think by letting people know that, for example, relationships between copper and iron and selenium and sodium and potassium are important — maybe there’s this music. Somebody cleverer than me can make something that can tune us in such a way that we are doing something that would help us.”