Cognitive Labs

Hope for early detection of Mild Cognitive Impairment may be around the corner with this new study.

Finnish scientists have identified metabolomic signatures for diagnosing patients with Alzheimer’s disease, as well as predicting which sufferers of mild cognitive impairment, or MCI, will go on to develop the disease.

Undertaken as part of the EU’s Predict AD research initiative, the project, which investigated markers in prospectively collected serum samples from more than 200 subjects, is one of the first large-scale efforts to discover metabolomic biomarkers for Alzheimer's, said Matej Oresic, a researcher at VTT Technical Research Centre of Finland and leader of the study.

Alzheimer’s disease is a significant area within proteomics research, with protein biomarkers seen as key to drug development, particularly with regard to selecting patient cohorts and monitoring therapy response during clinical trials (PM 12/2/2011). Indeed, in June a report commissioned by proteomics firm Proteome Sciences predicted that protein biomarkers for Alzheimer's disease will represent a cumulative $9 billion market over the next ten years (PM 6/3/2011).

Metabolomic biomarkers for Alzheimer’s have received significantly less attention, Oresic told ProteoMonitor, noting that “there are ongoing efforts, but not nearly as many as [there have been] in proteomics.”

While “there have been for some years [Alzheimer’s] studies looking at lipids — lipidomics, basically” — this work has “not been on a large scale, and developing a truly diagnostic model [using metabolomic markers] had not been done,” he said.

“Metabolomics is a relatively new field,” Oresic added. “I think people have been focused so much on proteomics for Alzheimer’s biomarkers because of the specific proteins [such as A-beta and tau] that have been so well studied” in models of the disease.

In the study, which was published in the current edition of Translational Psychiatry, the researchers used two-dimensional gas chromatography with time-of-flight mass spec on a Leco Pegasus 4D GC×GC-TOFMS instrument, as well as LC-MS on a Waters Acquity UPLC/Q-TOF Premier system to measure levels of 683 metabolites in 226 serum samples.

With this data, they performed a model selection via logistic regression analysis in multiple-cross validation runs to develop metabolite panels both for diagnosing Alzheimer’s and for predicting patient's progression from MCI to Alzheimer’s. The best diagnostic panel was selected in 248 of 1,000 cross-validation runs and consisted of three phosphatidylcholines along with ketovaline. The best panel for predicting progression was selected in 195 of 1,000 cross-validation runs and consisted of one phosphatidylcholine, an unidentified carboxylic acid, and 2,4-dihydroxybutanoic acid, which, Oresic noted, is responsible for the majority of this panel’s predictive power.

When combined with age, the diagnostic signature identified patients with Alzheimer’s with AUC of 0.81, with 67 percent sensitivity and 76 percent specificity. The predictive signature identified patients progressing from MCI to Alzheimer’s with AUC of 0.77, with 77 percent sensitivity and 70 percent specificity.

Protein biomarker studies led by University of Pennsylvania researchers Les Shaw and John Trojanowski have shown that measurements of A-beta and total tau protein levels in cerebrospinal fluid can predict progression from MCI to Alzheimer’s roughly 90 percent of the time (PM 6/11/2010) – better than the results obtained by the VTT team’s metabolomic panel. CSF is generally more difficult to obtain than serum, however, making desirable serum-based tests like that developed by Oresic’s team.

Adoption of CSF A-beta and tau as Alzheimer’s biomarkers has also been slowed by the difficulty of achieving good reproducibility across runs and across labs. Metabolomics approaches don’t eliminate this issue, Oresic said, but they do present a somewhat different set of challenges.

Mass spec-based proteomics has presented reproducibility issues in part due to the CVs associated with trypsin digestion and measurements of low-abundance analytes. Oresic said he was confident that standardizing mass spec-based quantitation of metabolomic biomarkers would not be a problem. Instead, he said, the challenge would come primarily from sample collection.

“In term of the quantitation, there shouldn’t be an issue [with reproducibility], but I think sampling is an issue,” he said. “Some metabolites are very sensitive to the treatment of samples — do they stay at room temperature or are they frozen immediately? Things like that.”

“That is something that would need to be investigated for each specific metabolite,” he added. “Some, like steroids, are very stable, but some are very sensitive. So the sampling and extraction part [of the workflow] really need to be carefully standardized and investigated to see how much variability we can expect.”

Metabolomic markers could prove useful in concert with proteomic markers, Oresic noted, a notion borne out, he suggested, by another Alzheimer’s study led by University of Gothenburg researcher Kaj Blennow and published contemporaneously in PLoS One.

That study, which Blennow undertook with researchers from biomarker detection firm Quanterix, used Quanterix’s Single Molecule Array technology to measure A-beta-42 levels in the serum of heart attack patients, finding that levels of that Alzheimer’s biomarker rose in subjects with severe hypoxia due to cardiac arrest.

The main biomarker in the VTT researchers’ predictive metabolomic profile — 2,4-dihydroxybutanoic acid — is also linked to hypoxia, Oresic noted. “They showed there is a link between hypoxia in the brain and what is going on with [A-beta-42] in the blood, and I think that is what we are picking up in our biomarkers, as well,” he said. “So it could be that in this early stage of Alzheimer’s lack of oxygen to the brain could be one of the initiating factors.”

“I think it’s still too early to tell, but there is potential for combing” metabolomic and proteomic markers, Oresic said. “Once we really know better what they are reflecting in terms of pathphysiology, it will be easier to think about the most optimal ways of combining them.”

He added that, “technologically, it shouldn’t be an issue,” noting that while in the Translational Psychiatry study his team used a GC/GC-TOF-MS instrument for measuring some small polar metabolites, this could also be done on an LC-MS system like those used for proteomics work.

The researchers now plan to run a validation study of their markers in a larger set of patients, as well as investigate them in CSF and biopsy samples from the original cohort, Oresic said. He added that he is involved in conversations with several large pharma firms regarding validation of the markers, but he declined to name any specific companies.

Labels: alzheimers, MCI, metabolomic, oresic

// posted by neurofuture9:30 AM permanent link