31 Mar 2017

Dying neurons release a slew of proteins into the brain and traces can make their way into the blood. A paper in the March 27 JAMA Neurology supports the idea that one of these proteins, neurofilament light (NfL), could be a biomarker for Alzheimer’s disease (AD). In the largest study yet on NfL in people with dementia, scientists led by Henrik Zetterberg and Kaj Blennow, University of Gothenburg, Sweden, and Niklas Mattsson, Lund University, Sweden, reported high levels of this protein in the blood of people with AD and mild cognitive impairment (MCI). Plasma NfL also associated with worsening cognitive scores over time and with brain atrophy. While it is not sensitive or specific enough to stand alone as a diagnostic marker, the protein distinguishes AD, MCI, and healthy controls about as well as do Aβ and tau in the cerebrospinal fluid (CSF), which is collected by spinal tap. Blood NfL would be a more easily accessible biomarker for prognosis and progression.

“The data are really strong,” said Anne Fagan, Washington University in St. Louis, who was not involved in the research. While the work needs to be replicated, a plasma marker that performs as well as ones in the CSF could have a huge impact, she told Alzforum. 

Climbing Concentrations. The average NfL found in plasma separates controls from MCI from AD. [©2017 American Medical Association. All rights reserved.]

NfL has been suggested as a marker of neurodegeneration in everything from amyotrophic lateral sclerosis to traumatic brain injury. Large, myelinated axons express the cytoskeletal protein and release it into the extracellular space when injured. A previous study from these Swedish researchers reported that NfL rises in the cerebrospinal of people with AD (see Nov 2015 news). Other studies suggested blood levels also climb in people with a variety of neurodegenerative disorders, including AD and atypical Parkinson’s (see Gaiottino et al., 2013; Jun 2016 news; Feb 2017 news). However, these studies included only a couple of dozen patients each and used ELISA-based assays that cannot detect less than about 78 pg/mL of NfL.

Blennow and Zetterberg’s lab developed a single-molecule array (Simoa) assay, which is 25 times more sensitive than ELISA, detecting as little as 0.6 pg/mL NfL in the blood. Mattsson and colleagues used this test on baseline blood samples from 180 patients with AD, 197 with MCI, and 193 cognitively healthy participants, all from the Alzheimer’s Disease Neuroimaging Initiative (ADNI). The researchers took advantage of the meticulous phenotyping of these volunteers, including their scores on various cognitive tests, markers of Aβ positivity, and structural imaging.

The researchers found that plasma NfL correlated with levels found in the CSF. While diagnostic groups strongly overlapped, average NfL was highest in people with AD (51.0 pg/mL), moderate in people with MCI (42.8 pg/mL), and lowest in cognitively normal people (34.7 pg/mL). MCI patients who were Aβ-positive by CSF analysis had more NfL than their Aβ-negative counterparts. Mattsson said too few clinically diagnosed AD patients were Aβ-negative for a robust comparison with those who were Aβ-positive, but said that all patients with a diagnosis of AD seemed to have similarly high NfL levels. Researchers have puzzled over patients with suspected non-Alzheimer pathology, and this NfL data would be another indication that they do have some form of neurodegeneration (see Aug 2016 news). 

Higher baseline NfL associated with worse baseline scores on the MMSE and ADAS-Cog, and with smaller hippocampus, thinner cortex, and greater ventricular volume as assessed by structural MRI. Elevated NfL also predicted decline in all measures over four years of follow-up, especially the MMSE, ADAS-COG 11, and cortical volume (see image below).

Predicting Progression: People in the highest quartile (Q4) of plasma NfL had the worst baseline MMSE and ADAS-COG II scores and declined more quickly. [©2017 American Medical Association. All rights reserved.]

Levels distinguished AD from controls with an area under the receiver operating characteristic curve, a statistical measure of assay specificity and sensitivity, of 0.87. This score is very similar to those for CSF Aβ (0.88), total tau (0.90), and hyperphosphorylated tau (0.87). It is rare to see such accuracy for a proposed AD blood biomarker, said Mattsson. His group is now searching for factors—such as ApoE status or other demographic information—that might be combined with NfL to improve diagnostic accuracy and help select people who should undergo more expensive or invasive testing. However, Mattsson said that multiple biomarkers, some more directly related to Aβ and tau, in addition to NfL would likely still be needed for an accurate AD diagnosis.

In an accompanying JAMA Neurology editorial, Douglas Galasko, University of California, San Diego, agreed with that assessment. “The idea that a single biomarker may reflect the complex pathogenesis of AD may be overly simplistic,” he wrote. “Perhaps a multianalyte panel of brain-derived biomarkers could form the basis for a plasma screening test, which might overcome the lack of specificity of a single biomarker alone.”

Given that blood NfL predicted disease progression, Mattsson suggested it could serve as a prognostic biomarker to help enrich clinical trial samples. It could also track neuronal injury in AD, which could be handy in longitudinal studies or drug trials that target neurodegeneration, the authors wrote. The group is planning to analyze ADNI blood samples to see if plasma NfL changes over time and how it tracks longitudinally with other markers of neurodegeneration.

The authors emphasized that NfL is far from specific for AD. They also pointed out that because ADNI excludes some people with underlying vascular burden, they may have missed an association of blood NfL with white matter hyperintensities, which has been described previously (Zetterberg et al., 2016).—Gwyneth Dickey Zakaib