Given the invasiveness and expense of measuring brain biomarkers for neurodegenerative diseases, researchers are searching for alternatives. Despite little success thus far, they continue to comb the complex jumble of proteins and molecules in the blood (see Aug 2014 conference story and May 2014 conference story). At the Alzheimer’s Association International Conference 2014, held July 12-17 in Copenhagen, Denmark, researchers presented their latest prospects, which ran the gamut from proteins and protein fragments to antibodies and aggregates.
Signs of Alzheimer’s disease
A previous study led by Samantha Burnham, Commonwealth Scientific and Industrial Research Organization (CSIRO), Perth, Australia, found that a panel incorporating five proteins in the blood predicted neocortical Aβ burden with a sensitivity and specificity of 80 and 82 percent, respectively, in the Australian Imaging, Biomarkers, and Lifestyle (AIBL) cohort (see Burnham et al., 2013). The panel included chemokine ligand 13, immunoglobulin M-1, pancreatic polypeptide (PPY), interleukin-17, and vascular cell adhesion protein (VCAM-1). At AAIC, Burnham presented follow-up research to suggest that this same panel predicted progression toward AD in 585 healthy controls and 49 people with mild cognitive impairment (MCI) followed over 54 months as part of the AIBL study.
Eleven percent of healthy controls with the amyloid-associated blood profile progressed to MCI or Alzheimer’s disease (AD), compared with 4 percent of controls without it. Eighty percent of MCI patients with this blood signature progressed to AD, compared with 29 percent of MCI patients without it. Those with the AD-like blood profile declined faster on tests of episodic memory, noted Burnham. “This suggests that a simple blood-based signature not only provides estimates of neocortical Aβ burden, but also identifies individuals at risk of progressing to AD at the prodromal and preclinical stages,” she wrote in an email to Alzforum. “With further validation we hope that this panel could provide a frontline screening tool for recruitment to clinical trials.”
Nicholas Ashton, working with Simon Lovestone from Kings College London, also used AIBL participants to look for blood evidence of neocortical amyloid. Rather than a panel of five markers, these researchers used an unbiased mass spectroscopy (MS) analysis to examine the serum of 40 people with high and 38 people with low amyloid burden as determined by PiB PET scans. They identified more than 4,500 unique peptides reflecting almost 1,000 isoforms of 380 different proteins. Of these, 69 isoforms correlated with positive PIB scans. Because MS tests are difficult to translate into routine clinical assays, Ashton narrowed these hits down to 17 proteins that might be suitable for ELISA testing. Three proteins significantly associated with amyloid burden on ELISA tests: α-2-macroglobulin, FHR-1, and Fibrinogen γ chain (FGG). Notably, those differed from the five proteins fingered by Burnham.
In a second cohort from the University of California, San Francisco, consisting of healthy controls and patients with MCI or AD—47 with high and 32 with low PiB uptake—only FGG predicted amyloid burden. The researchers reported at AAIC that together, age and blood FGG levels predicted whether a person had neocortical amyloid build-up with 59 percent sensitivity and 78 percent specificity. FGG is involved in inflammation and homeostasis and has previously been associated with amyloid plaques (see Liao et al., 2004). However, FGG was not among the 10 plasma biomarkers that this group recently reported to predict conversion from prodromal disease to AD (see Hye et al., 2014). The researchers will next validate FGG in a larger cohort, focusing on cognitively normal participants, said Ashton.
“For predicting who should get screened into clinical trials, these projects could be very useful,” said Sid O’Bryant, University of North Texas Health Science Center, Fort Worth. While both sensitivity and specificity are somewhat low, O’Bryant suggested that blood tests could help screen a potential clinical trial population for those most likely to have amyloid buildup. Researchers could then perform more expensive PET scans on only the people with a positive test, limiting the cost of trial recruitment.
Other researchers are looking at antibodies that might signal disease. Robert Nagele, Rowan University, Stratford, New Jersey, explained that when a neuron dies, it releases molecules into the cerebrospinal fluid (CSF) that make their way into the blood. To clear the debris, the immune system produces autoantibodies, some of which may be specific to particular neuronal populations, Nagele said. He detects these serum autoantibodies by capturing them on a microarray of almost 9,500 human proteins.
Previously, Nagele reported that a panel of 10 autoantibodies distinguished commercially supplied blood of mild to moderate Alzheimer’s patients from that of controls with 96 percent sensitivity and 92.5 percent specificity (see Aug 2011 news story). At AAIC, he talked about extending the method to an earlier stage of AD and to other diseases.
Nagele examined blood samples from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). His panel separated 50 MCI patients whose low CSF Aβ42 put them at high risk of progressing to Alzheimer’s dementia, from 111 controls with a sensitivity and specificity of 98 percent each. These 10 autoantibodies were different from those he reported previously to detect mild to moderate AD, suggesting that there may be specific antibody signatures for different stages of the disease, Nagele said. Interestingly, the panel did not include antibodies against the Aβ peptide. It included an inflammatory marker, two microtubule-associated proteins, a growth factor regulator, an apoptotic regulator, and five uncharacterized proteins. Nagele plans to validate his MCI test in additional ADNI samples, especially from participants who enrolled as controls and later developed AD. Their archived blood samples would help determine whether this test can detect disease in presymptomatic individuals, he said.
Dave Morgan, University of South Florida, Tampa, praised the approach because it involved no a priori hypothesis, but instead looked widely for a signal that distinguished healthy people from others with AD or Parkinson’s disease (see below). Morgan sees a scientific rationale for picking up different autoantibody signatures in people with disease; however, the tests will need to be validated in much larger cohorts.
Nagele developed a separate autoantibody biomarker panel to distinguish people with early stage Parkinson’s disease from healthy controls. It correctly identified 89.2 percent of patients, he reported at AAIC. For clinicians looking for signs of neurodegeneration in general, Nagele’s group also developed a 50-marker panel that detected factors common to AD, PD, and multiple sclerosis. It differentiated disease from control samples with an error rate of 10.5 percent. Nagele said such a test could eventually serve as a general screen for neuronal damage, but it, too, needs validation in larger cohorts.
“The error rates seem quite reasonable,” said Morgan. If a blood test picked out people at risk for pathology with even 20 percent error, it would cut down on the number of more definitive, but expensive and invasive, tests used in people who have no pathology, he said. That said, since clinical diagnoses are wrong about 20 percent of the time, Morgan cautioned that Nagele’s actual error rate may be higher than calculated.
Blood Signature of Neurodegeneration
Other researchers are also looking for signs of general neurodegeneration. Dilek Inekci, working with Kim Henriksen of Nordic Bioscience, Herlev, Denmark, is pursuing tests for tau, which aggregates in several neurodegenerative disorders. While intact tau is difficult to detect in the blood, its fragments cross the blood-brain barrier more easily and some scientists think they could be a marker of disease. Inekci explained in a talk that after a neuron enters apoptosis, caspase-3 cleaves at the aspartate-421 site of tau and generates a C-terminal fragment called tau-C.
The researchers generated a monoclonal antibody specific for this fragment, which did not react with the intact form of tau. They used the antibody to immunoprecipitate the fragment from the serum of two AD patients and two healthy controls, and identified the immunoprecipitation products by mass spectrometry. While none appeared in the serum of normal controls, the fragment turned up in samples from AD patients. Quantifying the protein levels using ELISA, AD patients had twice as much tau-C as healthy controls.
Using the same antibody, the researchers developed an ELISA to detect tau-C and tried it on blood samples of eight people with hypoxic brain injury due to cardiac arrest. Patients had 10-fold more of the fragment than healthy controls. The researchers then performed the ELISA on blood from 18 people with AD, 18 with another type of dementia, and 18 controls. AD patients showed a non-significant trend of higher tau-C levels compared to the other two groups. “We don’t expect the fragment to be Alzheimer’s-specific, since you can see it rise in other cases of neuronal damage,” said Inekci. “Rather, it’s a marker of neuronal damage.” Such a marker could be used to predict concussion outcomes, Henriksen said. Inekci presented a separate poster reporting that a higher amount of tau-C reflected poor results following stroke in 19 patients.
Damian Crowther, University of Cambridge, U.K., is looking for a way to detect neurodegeneration in the blood by exploiting the tendency of many disease-related proteins to seed aggregation of normal forms. He and colleagues have developed a microfluidic approach to detecting seeds in brain and serum. The method involves loading a serum sample, fluorescently labeled synthetic Aβ42 monomers, and molten agarose onto a 1-centimeter plastic chip containing 10-50,000 picoliter wells. After incubating for three hours, the agarose cools into a gel, and researchers wash away unaggregated monomer. Any clumps greater than 1,000 kilodaltons stay behind. The researchers then look to see if any of the wells contain a fluorescing bundle. That would indicate that a seed was present in that well, i.e., the serum.
Crowther and colleagues applied this technique to brain extracts from flies and mice. It detected seeds in Drosophila that express Aβ42, but not in control brains. Likewise, the assay detected Aβ seeds in CRND8 mice, which overexpress a mutant version of human APP and overproduce Aβ42. It found seeds in serum samples from mice that were seven months old, an age at which they are robustly accumulating plaques. Crowther said he plans to examine how the test works in human serum samples. “This could be a good screen for drug candidates that stop the seeding process,” he told Alzforum.
One audience member asked if this seeding test works with tau or α-synuclein monomers. Crowther has not tried that but said it might be possible to multiplex this assay, where differently colored fluorescent tags added to a single sample label different types of monomer.—Gwyneth Dickey Zakaib
- Exosomes Stand Out as Potential Blood Biomarkers
- Blood Patterns: Prowling for Alzheimer’s Clues in Plasma
- "Autoantibody-omics" Yields Potential Blood Biomarkers for AD
Research Models Citations
- Burnham SC, Faux NG, Wilson W, Laws SM, Ames D, Bedo J, Bush AI, Doecke JD, Ellis KA, Head R, Jones G, Kiiveri H, Martins RN, Rembach A, Rowe CC, Salvado O, Macaulay SL, Masters CL, Villemagne VL, . A blood-based predictor for neocortical Aβ burden in Alzheimer's disease: results from the AIBL study. Mol Psychiatry. 2013 Apr 30; PubMed.
- Liao L, Cheng D, Wang J, Duong DM, Losik TG, Gearing M, Rees HD, Lah JJ, Levey AI, Peng J. Proteomic characterization of postmortem amyloid plaques isolated by laser capture microdissection. J Biol Chem. 2004 Aug 27;279(35):37061-8. PubMed.
- Hye A, Riddoch-Contreras J, Baird AL, Ashton NJ, Bazenet C, Leung R, Westman E, Simmons A, Dobson R, Sattlecker M, Lupton M, Lunnon K, Keohane A, Ward M, Pike I, Zucht HD, Pepin D, Zheng W, Tunnicliffe A, Richardson J, Gauthier S, Soininen H, Kłoszewska I, Mecocci P, Tsolaki M, Vellas B, Lovestone S. Plasma proteins predict conversion to dementia from prodromal disease. Alzheimers Dement. 2014 Nov;10(6):799-807.e2. Epub 2014 Jul 8 PubMed.