(Not yet) a blood test to detect preclinical Alzheimer’s disease

Image courtesy of the National Institute on Aging/National Institutes of Health.

A team of US researchers recently reported in Nature Medicine having identified a panel of ten molecules whose concentration in blood could predict whether someone in the group of people studied would go on to develop signs of Alzheimer’s disease within the next two or three years or not.

Does this mean that it will soon be possible to tell people whether they will develop Alzheimer’s disease or not from a simple blood test, as some headlines in the media seem to imply? Not quite. To understand why, we need to look a bit more closely at the study, its results and limitations.

  • What are the results of the study?

The researchers conducted a 5-year clinical study in which they enrolled a total of 525 healthy individuals aged 70 or older. Participants had to be free of known major psychiatric or neurological illnesses (including diagnosed Alzheimer’s disease) to enter the study. They were then followed prospectively with yearly cognitive assessments and collections of blood samples.

Discovery cohort

Since the enrollment in the study was done on a rolling basis (meaning that not all individuals were enrolled at the same time on day 1 of the study, but rather enrolled continually over time), and since there were drop-outs (people who are enrolled in the study but stop participating at some point for whatever reason), three years after the official start of the study there were 467 people who had completed the baseline assessment (first set of cognitive measures and blood test), 394 who had completed baseline and 1-year follow-up, and 202 who had completed baseline, 1-year follow-up and 2-year follow-up. The researchers had therefore at that time 202 participants at their disposal for analysis.

The team divided these 202 participants into three groups, based on the results from the cognitive assessment done at the 2-year follow-up visit:
– 53 people had either early Alzheimer’s disease (AD) or amnestic mild cognitive impairment (aMCI, a memory impairment which often progresses to AD),
– 96 people were defined as control subjects (unimpaired cognitive function),
– 53 people did not meet the criteria for either of these two groups and were therefore not included in the following analysis.

Of the 53 people with either AD or aMCI, 35 were actually “incident” cases, meaning that although they had never been diagnosed with AD or cognitive impairment prior to enrollment in the clinical study, measurements at baseline had revealed that they in fact had either early AD or aMCI. The other 18 cases were called “converters” because they had normal results in the cognitive assessments at baseline but developed signs of aMCI or early AD during the study (detected at one of the follow-up tests).

The researchers randomly selected 53 people from the 96 normal controls (matching them to the 53 aMCI/AD cases for similar age, education level, and sex) and then analyzed molecules present in the blood of all these people (106 individuals in total), using the most recent blood samples available for the “incident” aMCI/AD group and control group and blood samples from both baseline and 2-year follow-up for the “converters” (that is, blood samples collected before and after appearance of symptoms).

Analysis of molecules present in the blood samples identified 10 lipids for which blood levels were on average lower in the 18 “converters” at baseline (when they had yet no signs of cognitive impairment) than in the control group. Those levels remained low at the 2-year follow-up (when the converters had developed signs of aMCI/AD) and were similar to the levels found in the blood of the aMCI/AD group (“incident” cases). The researchers therefore hypothesized that this set of 10 lipids may help identify people with normal cognitive function who would go on to develop signs of memory impairment within 2-3 years.

Validation cohort

The research team then analyzed the levels of the lipids they had identified in the discovery phase in a set of 40 other individuals to see if they could confirm the results obtained in the discovery cohort.

Why only 40 individuals?

At the end of the 5th year of the study, there were a total of 525 people who had completed baseline assessments but only 295 people who had completed three consecutive visits (baseline, 1-year and 2-year follow-ups) and who had not already been included in the discovery phase analysis. Out of these 295 people,
– 124 were classified as controls,
– 21 as aMCI/AD (out of which 11 were “incident” cases and 10 “converters”),
– 150 did not meet the criteria to be classified as either control or aMCI/AD and were therefore not included in the following analysis.

As in the discovery phase, 20 individuals from the control group were selected and matched to the 20 aMCI/AD individuals analyzed (apparently one aMCI/AD case was not available for analysis) and blood lipid levels were measured. The researchers observed that the levels of the 10 lipids they had singled out in the discovery phase were lower in the converters than in the controls even before the converters showed any sign of memory impairment, and that the levels were comparable between converters before appearance of symptoms and established aMCI/AD cases.

  • What can be said from these results?


To recap briefly, the researchers showed that, in their study group, there was a set of 10 lipids for which the concentration in blood differed between people who had normal cognitive function but went on to develop memory impairment within two to three years and people who had no signs of memory impairment. Using these data, the researchers built a model based on the blood levels of these 10 lipids that could effectively predict who in their study group would go on from normal cognitive function to develop aMCI/AD (memory impairment) within 2-3 years. The authors of the study report 90% sensitivity and 90% specificity for their model.


The key words one should notice here are “in their study group”. The authors of the study as well as the rest of the scientific community who commented on the study are indeed well aware that the results need to be replicated in larger and more diverse groups of people before anything remotely close to “blood test for detecting preclinical Alzheimer’s disease” can be claimed (remember all the inclusion/exclusion/classification criteria to be fulfilled at different steps in the study, and the small number of individuals who were included for analysis in the end).

Researchers have also to verify that the panel of lipids identified in this study is indeed specific for detecting early Alzheimer’s disease versus other forms of cognitive impairment. A limitation of the study is indeed that the classification of people as having early Alzheimer’s disease was based on cognitive tests and not confirmed by standard cerebrospinal fluid tests or PET imaging scans.


What then makes this study interesting, even if it’s not yet sure that its results will hold true in other clinical trials? If nothing else, it presents researchers with a list of ten molecules to look at more closely. If the results hold true in other study groups, then great, scientists will be a step closer to develop a blood test to detect preclinical Alzheimer’s disease. If they don’t, scientists will at least know what does not work, and that in itself is a gain of knowledge (though, it is true, a gain that is not as valued as positive results by academic journals and scoop-hungry general media alike).

  • To sum up

Headlines in the media may have given the general impression that researchers had just developed a screening test to predict whether one would develop Alzheimer’s disease or not. After looking at the original research paper presenting the study and the data it contains, what can really be said?

– No, researchers have not just developed a blood test that can tell anyone whether they will get Alzheimer’s disease within the next few years or not.

– Yes, researchers have identified a set of ten molecules which, in their particular study group that satisfies a particular set of criteria, can distinguish between people who will develop memory cognitive impairment within 2-3 years and people without cognitive impairment.

These ten molecules are now worth investigating further as a next step on the multi-step trial-and-error process that is the search for a way to detect Alzheimer’s disease before symptoms appear and disease is already well advanced. Now, of course, whether it is desirable to be able to detect that someone will develop a certain disease in the future when there is currently no known way of preventing said disease or of curing it remains a matter of debate.


* Considering the number of news articles in the general media stating something like “A new blood test can predict Alzheimer’s disease with 90% accuracy”, and how such a statement can mislead the public as to what the true predictive power of the test would be, I have now written another post to deal with this particular issue.

Also on the topic:
Diagnosing dementia – This is not spinal tap. The Economist, March 15th 2014.

Plasma phospholipids identify antecedent memory impairment in older adults. Mapstone M, Cheema AK, Fiandaca MS, Zhong X, Mhyre TR, Macarthur LH, Hall WJ, Fisher SG, Peterson DR, Haley JM, Nazar MD, Rich SA, Berlau DJ, Peltz CB, Tan MT, Kawas CH, Federoff HJ. Nat Med. 2014 Mar 9. doi: 10.1038/nm.3466
PMID: 24608097

ResearchBlogging.orgMapstone M, Cheema AK, Fiandaca MS, Zhong X, Mhyre TR, Macarthur LH, Hall WJ, Fisher SG, Peterson DR, Haley JM, Nazar MD, Rich SA, Berlau DJ, Peltz CB, Tan MT, Kawas CH, & Federoff HJ (2014). Plasma phospholipids identify antecedent memory impairment in older adults. Nature medicine PMID: 24608097


One thought on “(Not yet) a blood test to detect preclinical Alzheimer’s disease

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google+ photo

You are commenting using your Google+ account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )


Connecting to %s