Jing Zhang is part of a group of researchers who published a study in the August edition of Journal of Alzheimer’s Disease on finding biomarkers for neurodegenerative diseases such as Alzheimer’s, Parkinson’s, and dementia with Lewy body disease.
The researchers used a multiplex quantitative proteomics method, iTRAQ [isobaric tagging for relative and absolute protein quantification], in conjunction with multidimensional chromatography and tandem mass spectrometry to study changes in the proteome of cerebrospinal fluid in patients with the diseases. They then compared those changes to healthy controls.
The study also included researchers from Oregon Health and Science University in Portland; Baylor College of Medicine in Houston; the Fred Hutchinson Cancer Research Center in Seattle; and Applied Biosystems.
ProteoMonitorcaught up with Zhang to talk about his work.
Can you sum up what you did in the study, and your findings?
The basic problem we have is that we have no biomarkers that are available [for neurodegenerative diseases]. There are no established biomarkers to assist the clinical diagnosis of Alzheimer’s, Parkinson’s, or dementia with Lewy body disease. Clinical diagnostic accuracy of those diseases depends on the disease. For Alzheimer’s it can get up to 90 percent. Parkinson’s, depending on the institution, people usually quote 78, 80 percent. If you go to dementia with Lewy body disease, you’re essentially flipping a coin.
For clinical discovery — proteomic discovery, metabolic discovery, whatever, genomic discovery — people have done all sorts of things. None have been very successful. On the proteomic end, this is the first study that compared multiple diseases simultaneously. In the past people have done Alzheimer’s versus control, Parkinson’s versus control, so on and so forth. But this is the one that compared multiple diseases simultaneously with iTRAQ.
The advantage of that is that there is a major caveat with proteomic discoveries. That is, if you don’t see a protein in one setting, you don’t know whether it’s not there or if it’s because of a variation of the technology. So this gets around that.
This is the fundamental concept before we started our experiment. There are several important aspects of this study that need to be emphasized. Number one is that patients with Alzheimer’s disease, dementia with Lewy body disease, they areall pathologically verified. That is a tremendous advantage over other studies.
So what we did was we collected cerebrospinal fluid from patients before they died and then … [when they died] we did the pathology confirmation [on the brain]. [For] those with confirmation … we went back to the archives of CSF to do the analysis.
The other thing we did was an extensive profiling of CSF. In this particular study, we have gone through more than 1,500 proteins. If we saw one protein in Alzheimer’s disease, we looked specifically for this protein in other groups automatically. This is not something that can be achieved if you just do cataloguing of proteins.
By iTRAQ technology, we saw many hundreds of proteins. Among the 1,500 proteins, we identified more than 300 proteins that changed over control. So if you look at this group alone, [you might say] ‘Aha, I discovered more than 300 proteins that are unique to Alzheimer’s disease.’ But that’s not true. For all the proteins that people claim to be specific to Alzheimer’s disease, very few of them actually are unique to Alzheimer’s disease because those proteins also are found in other diseases including Parkinson’s and dementia with Lewy body disease.
By multi-comparison, we isolated proteins that can only be found in Alzheimer’s disease, Parkinson’s disease, and dementia with Lewy body disease.
And finally we went even further. We bought a whole bunch of antibodies and then validated a subset of those proteins. At the end of the day we actually discovered several panels of biomarkers that can segregate Alzheimer’s or Parkinson’s or DLB from control as well as from each other.
Why has it been so difficult to find established biomarkers for these diseases?
One reason is the nature of the diseases. People always say Alzheimer’s disease, Parkinson’s disease. They probably don’t mean it’s a disease. It’s a syndrome. In other words, there’s heterogeneity to each disease. Each subgroup of people, they can actually exhibit different markers. So people in the past, they tried looking for a bullet marker, which is not possible in my eyes.
Different people exhibit different markers because they’re not single diseases. It’s a syndrome. Some people with Alzheimer’s have hallucinations, some don’t. There’s a whole host of symptomology. Same thing with Parkinson’s. Some people predominate with tremors, some people predominate with rigidity. It all depends on the particular patient you study. There are different markers expressed.
So the most powerful way [of trying to predict disease] is the combination of markers. It’s the combination of markers that can achieve high sensitivity and high specificity.
Another issue is technology. Before we introduced the LC/MS/MS method into the biomarker market, most people had been working on CSF biomarkers by two-dimensional gel. Very few people have done pre-fractionation. In other words, all they’ve done is with 100, 200 microliters of protein loaded on 2D gel, and see which one has changed versus a control.
It’s well known that 2D gel has a resolution, a dynamic range of about 104. But the CSF can do 109 minimum. So if you think about it, you can load the protein on a 2D gel, what can you see? Only really abundant proteins.. But the CSF can do 10minimum. So if you think about it, you can load the protein on a 2D gel, what can you see? Only really abundant proteins.
What’s the significance of your study and how does it move the study of such diseases forward?
We’re in the process of validating those markers in the larger population, more than 200 patients. If we can validate them, the next step would be the usage of those markers. They can be used for the diagnosis of those diseases, tracking disease progression and assessing therapeutic effects of pharmaceutical drugs.
The newer markers, they could be potentially therapeutic targets.
How is what you’re doing different from people who are studying these diseases on the genetic level?
The genes are not a good way of doing biomarker discovery. Gene studies usually need isolation of tissues of interest, in this case, the human brain. It is not really possible to obtain brain [tissue] before [the] death of a patient with neurodegenerative diseases. The alternative could be to collect white cells in the blood of patients for gene studies. However, unless the disease is with a defined genetic problem — neither sporadic Alzheimer’s nor Parkinson’s can qualify in this regard — the chances for getting a marker specifically related to the brain in peripheral white cells is exceedingly low. In addition, peripheral markers are vulnerable to confounding factors such as diseases unrelated to the brain such as cardiovascular problems and diabetes, as well as the medicine used to treat these nonspecific diseases.
So if [researchers] use the methods you used in your study, is it going to be possible to identify people predisposed to Alzheimer’s before they show symptoms of it?
Actually, that’s the direction we’re moving toward: preclinical diagnosis. But how do you identify preclinical patients? That’s the issue. So we’re collaborating with a whole bunch of people using neuroimaging and studying patients according to their MCI, mind cognitive impairment. Then we follow them. A subset of them will develop Alzheimer’s. Then we compare the CSF, the A setting versus the B setting. And then we can get the preclinical markers. That’s one way of doing it.
Another way of doing it is to use our current markers to test whether those markers are also present in the MCI patients and then follow them to see which ones will develop Alzheimers’s and see how our markers perform there.
It sounds like with what you’re doing, we’re at the very, very beginning stage of Alzheimer’s research.
It depends on the purpose. If you’re talking about diagnosing disease, already it’s very close. If you’re talking about predicting disease, who will develop Alzheimer’s, yes, it’s very early.