Current models of biomarker change in Alzheimer’s disease have been based mostly on cross-sectional data. In the March 5 Science Translational Medicine, researchers led by Anne Fagan at Washington University in St. Louis provide the first glimpse of longitudinal data from families enrolled in the Dominantly Inherited Alzheimer Network (DIAN). Unexpectedly, the data trajectory differed in some respects from that predicted by the models. Once DIAN participants showed symptoms, cerebrospinal fluid (CSF) markers of neurodegeneration, such as tau, dropped slightly over time. This contradicts findings from cross-sectional studies, which consistently measure higher amounts of CSF tau in people at later stages of the disease. Fagan noted that the results will need to be confirmed in much larger groups, as the initial longitudinal sample was small. DIAN participants carry rare gene mutations that cause AD in middle age, leaving it unclear whether the findings will generalize to people with late-onset AD. Nevertheless, the results highlight the importance of gathering longitudinal data to refine biomarker models. “Comparing people cross-sectionally misses some important information,” Fagan said. The data could have implications for researchers using biomarkers as measures of efficacy in clinical trials. 

“This is an important contribution that sheds some light on what is happening dynamically,” said Ronald Petersen at the Mayo Clinic in Rochester, Minnesota. “Not infrequently, longitudinal data do not bear out cross-sectional data. It points to the importance of gathering longitudinal data to guide the design of trials.”

The classical model proposed by Clifford Jack at the Rochester Mayo Clinic predicts that Aβ markers first become abnormal 20 years or more before symptoms appear, followed by CSF tau and other measures such as brain volume as seen by MRI (see Jan 2010 webinar). That model was refined in 2013 to track biomarkers by time rather than disease stage, which allowed for the fact that people with similar biomarker profiles often have different clinical symptoms (see Feb 2013 conference story). Cross-sectional data from studies of familial Alzheimer’s disease strongly support the updated model, suggesting that familial and sporadic forms of the disease progress similarly (see Jul 2012 news storyAug 2012 conference storyNov 2012 news story).

In the new paper, DIAN researchers report longitudinal data from 26 AD mutation carriers and 11 controls, each of whom provided two or more CSF samples over a period of one to three years. The findings varied depending on where each person was in the disease course. In people who were younger than their expected age of disease onset, CSF tau rose over time. However, in people past their age of expected onset, tau, phosphorylated tau, and visinin-like protein-1 (VILIP-1) all fell slightly (about 10 pg/ml) but significantly over time. All three of these proteins are considered markers of neuronal death, as they spike after various forms of brain damage such as stroke or traumatic brain injury. VILIP-1 is a calcium sensor that, like tau, resides inside neurons and only appears in the extracellular space after damage, and is being tested as a tau-independent injury marker (see Mar 2012 news story). While these CSF findings imply that tau switches from rising to falling around the time of disease onset, the longitudinal sample does not yet include any data from a person before and after this transition. As more participants develop symptoms, the picture should become clearer, Fagan said. 

Why would these injury markers fall after disease onset? Fagan speculated that neurodegeneration could be slowing down at this stage, as many of the at-risk neurons have already died. According to MRI studies, the brain continues to shrink as AD advances, but ongoing atrophy does not necessarily conflict with declining levels of injury markers, Fagan said. Tau release may precede the later stages of neurodegeneration when cells wither away entirely, for example. In future longitudinal studies, she plans to correlate changes in CSF markers with MRI and tau imaging to better understand the relationships among them.

“The decline in markers of neurodegeneration was a very intriguing finding, and the authors provided plausible hypotheses to explain it,” said Leslie Shaw at the University of Pennsylvania School of Medicine, Philadelphia. Shaw runs the biomarker core at the Alzheimer’s Disease Neuroimaging Initiative along with John Trojanowski. Shaw noted that the DIAN data dovetail with some of ADNI’s recent longitudinal findings, which show falling CSF tau in sporadic AD patients (see Toledo et al., 2013). Likewise, a longitudinal study in Finland found that CSF p-tau declined over a period of about three years in subjects with AD, but rose in people at earlier stages of the disease (see Seppälä et al., 2011). These data hint that injury markers may behave similarly in sporadic and familial disease, but more studies are needed to prove this, Shaw said. He pointed out that sporadic disease strikes later in life and usually includes additional pathologies such as vascular disease or α-synuclein deposits, which may affect the rates at which tau and Aβ accumulate (see Jul 2013 news story). 

In addition to the longitudinal data, Fagan and colleagues extended their cross-sectional results in the new paper. They had previously published baseline data from the first 128 DIAN participants, and now have nearly doubled that number to 146 mutation carriers and 96 non-carriers. As before, they saw the highest CSF Aβ42 in carriers 25 years before the expected age of disease onset. At that point, carriers had more Aβ42 than controls did. The marker stayed constant in controls, but levels were lower in mutation carriers at every later stage, up to 10 years after onset. Aβ42 first dipped significantly below control levels at 10 years before onset. Tau, p-tau, and VILIP-1 became significantly abnormal in people 15 to 20 years prior to symptoms, and were higher, on average, at every subsequent age up to 10 years after onset. Because this data did not track what happened in individuals over time, it missed the slight decline in injury markers seen in the longitudinal data.

Henrik Zetterberg at the University of Gothenburg, Sweden, found it intriguing that injury markers seemed to change in parallel with Aβ in this cohort, rather than following amyloid changes by several years as the Jack model proposes. “It makes better sense from a biological standpoint for tau to be released when amyloid starts to injure neurons, rather than following it by five or 10 years,” he told Alzforum. The Biomarkers in Older Controls at Risk for Dementia (BIOCARD) study found similar parallel movements in Aβ and tau, Zetterberg noted (see Oct 2013 news story). Indeed, the latest revision to the Jack model allows for early, low-level tau accumulation that shoots up after being exacerbated by Aβ (see Feb 2013 conference story). 

Many researchers wondered what effect the new findings could have on the use of biomarkers as readouts in clinical trials. If injury markers do, in fact, fall in people with symptomatic AD, how would a successful treatment be expected to affect them? Would CSF tau fall even more steeply, or would it rise, because people would in effect be held earlier in the disease course, when tau was higher? At this point, no one knows. “It does give you pause to ask, do we know enough about these biomarkers and their longitudinal performance to tell if a drug is working or not working?” Petersen said.—Madolyn Bowman Rogers.


Make a Comment

To make a comment you must login or register.

Comments on this content

No Available Comments


Webinar Citations

  1. Together at Last, Top Five Biomarkers Model Stages of AD

News Citations

  1. HAI—Sharper Curves: Revamping a Biomarker Staging Model
  2. Paper Alert: DIAN Biomarker Data Show Changes Decades Before AD
  3. In Big Picture, Familial AD’s Biomarker Data Resemble LOAD
  4. API Echoes DIAN: Biomarker Changes Precede Symptoms by 20 Years
  5. Research Brief: VILIP-1 a Potential CSF Marker for AD?
  6. An Extra Strain on the Brain—α-Synuclein Seeds Tau Aggregation
  7. Speed of Biomarker Changes Herald Alzheimer’s Disease

Paper Citations

  1. . Longitudinal change in CSF Tau and Aβ biomarkers for up to 48 months in ADNI. Acta Neuropathol. 2013 Jun 29; PubMed.
  2. . Longitudinal changes of CSF biomarkers in Alzheimer's disease. J Alzheimers Dis. 2011;25(4):583-94. PubMed.

External Citations

  1. Alzheimer’s Disease Neuroimaging Initiative

Further Reading

Primary Papers

  1. . Longitudinal change in CSF biomarkers in autosomal-dominant Alzheimer's disease. Sci Transl Med. 2014 Mar 5;6(226):226ra30. PubMed.