Introduction
Trisomy of chromosome 21, commonly known as Down syndrome (DS), is the most frequent genetic cause of lifelong intellectual disability and occurs in over 7 million people worldwide [
1]. DS is also the most common genetic cause of early onset Alzheimer’s disease (AD) [
2], and in prospective studies, the cumulative incidence for dementia is around 90% [
3‐
5]. Given that the AD-associated
APP gene is on chromosome 21, and that rare DS individuals who lack a third copy of
APP do not develop AD [
6], overexpression of APP and accumulation of the amyloid β-protein (Aβ) are considered the main drivers of AD in DS.
Early clinical diagnosis of AD in people with DS is complex and challenging because patients with DS have a pre-existing intellectual disability, the extent of which varies from person to person. Reliable AD-specific biomarkers would greatly assist early diagnosis of AD in DS, and could facilitate clinical care planning, and timely treatment with disease-modifying AD therapies which may soon become available [
7,
8]. Measurement of Aβ and tau by their detection in CSF using immunoassays or visualization of deposited proteins using PET imaging can reliably detect AD in at-risk individuals [
9‐
11]. However, PET imaging is expensive and its application outside of clinical research remains limited. Collection of CSF, although routine in certain countries, remains unpopular with patients, especially if required more than once. Lumbar puncture and PET scans are particularly challenging in a vulnerable population such as people with DS. In contrast, collection of blood would be significantly easier for individuals with DS, making blood-based biomarkers of AD in DS the preferred option.
Postmortem studies of brains from individuals with DS reveal that in general, diffuse amyloid plaques appear in the late teens, and tau pathology emerges after age 35 [
12,
13]. Whether blood-based biomarkers of AD evolve in a similar temporal pattern is not yet clear. Prior studies of putative AD biomarkers examined restricted age groups [
14‐
16], and earlier studies employed sub-optimal assays [
17]. Most previous investigations on AD biomarkers in DS were limited to analysis of plasma Aβ. Now with the advent of reliable methods, it is possible to assess whether tau and neurofilament light (NfL) are also altered in DS. NfL, a scaffolding cytoskeleton protein, is elevated in plasma in many neurodegenerative conditions [
18], and several NfL assays are available [
19]. Measurement of tau in plasma has been less straightforward. This is because tau is molecularly heterogeneous [
20‐
22] and is present in blood at only minute levels. We recently developed an ultra-sensitive immunoassay, which detects forms of tau captured by the mid-region antibody BT2 (to aa 194–198) and detected with the N-terminal antibody Tau12 (to aa 6–13). In several studies, we have found that measurement of tau using this NT1 assay effectively discriminates AD from controls [
21] and is elevated in patients with mild cognitive impairment who subsequently developed AD [
23]. Other assays which target distinct epitopes of tau have been reported, but these have shown less consistent differences between AD and controls [
24‐
26].
Here, we measured Aβ42, NT1 tau, and NfL in plasma from individuals with DS and age-matched controls. Our study cohort covered a broad age range, including participants as young as 3 months and as old as 68 years. The study had 2 primary objectives: (1) to assess changes of plasma Aβ42, NT1 tau, and NfL in DS across age, and (2) to compare biomarkers measured in DS plasma versus age- and sex-matched controls. The usefulness of any biomarker is influenced by a myriad of factors, key among which are the stability of the analyte(s) over time, e.g., day-to-day variability, and whether or not it is affected by diurnal factors and requires pre-sampling fasting. Prior to analyzing precious specimens from our unique DS-control matched cohort, we examined the in vivo stability of analytes in the plasma of 10 healthy volunteers collected at 6 time points over a 5 day interval.
Discussion
Here, we analyzed changes in biomarkers related to 3 primary features of AD and AD in DS: amyloid, tau, and neurodegeneration [
29]. In the absence of large longitudinal cohorts, we gathered specimens from 100 DS individuals aged from 3 months to 68 years and compared their values with those of age- and sex-matched controls. We found that Aβ42 levels were higher in DS than in controls, regardless of age. In both DS and controls, Aβ42 levels were highest in neonates. NT1 tau levels were similar in DS and controls across all ages, except for older ages when NT1 tau was higher in DS than controls. For both DS and controls, NfL levels were relatively low in age groups up to ~ 30 years, whereas in older age groups, NfL was higher and the increase was greater in DS than in controls.
As expected for individuals with 3 copies of the
APP gene (and consistent with prior studies [
17,
30‐
32]), we observed higher plasma Aβ42 levels in individuals with DS compared to controls. In the first decade of life, a time when there is little amyloid deposition [
33,
34], plasma Aβ42 levels were on average a little higher (~ 1.6 fold) in DS than the expected 1.5-fold elevation due to gene dosage (
p = 0.03). Why this should be is unclear, but it is worth noting that several AD risk factors are encoded on chromosome 21 [
2] and these might contribute to either enhanced amyloidogenic processing of APP [
35,
36] or reduced degradation of Aβ [
37]. Also, neuronal Aβ production is activity dependent [
38], and in DS, there is evidence of aberrant hyperactivity during development and early life [
39] that could contribute to higher Aβ levels.
Aβ42 levels tended to fall with age in both individuals with DS and controls, but the relative decrease was greater in DS. There was a strong tendency for the DS/control Aβ42 ratio to be lower in the oldest group (> 50 years) compared to youngest group (0–10 years); however, this did not reach statistical significance (1.6 vs. 1.4,
p = 0.08). Nonetheless, the trend is consistent with the notion that Aβ42 is prone to aggregate and becomes trapped in accumulating plaques and is in line with multiple AD studies linking falling CSF and plasma Aβ42 with increased cerebral amyloid deposition [
40,
41]. However, prior studies examining the association of plasma Aβ42 with age in DS have yielded conflicting results with reports of increased [
42], decreased [
43], and unchanged Aβ42 levels [
14,
31,
44]. But previous studies were not designed to look at the effect of age across a broad age span. We found that plasma Aβ42 levels fell sharply in the first 3 decades of life in individuals with DS, but were relatively stable in the age range from 31 to 68 years. Our results from a broad age range of individuals (3 months to 68 years) provide the perspective to better understand what had formerly appeared discordant results, that is, decreasing Aβ42 levels in younger DS individuals [
43] but relatively stable levels in older DS individuals [
14,
31,
44].
Human plasma is a complex matrix, components of which can interfere with immunoassays. One means of overcoming matrix interference is to dilute samples so as to reduce interfering plasma components to a level below which they no longer interfere. This requires that the assays employed are sufficiently sensitive to allow dilutions necessary to preclude matrix interference. Here, we employed ultra-sensitive assays and evaluated the maximum dilution that allowed consistent detection of analytes across a large number of human samples. For Aβ42, only a few studies [
27,
32,
45,
46] have used such ultra-sensitive techniques.
In contrast to Aβ, only a handful of studies have attempted to measure tau in plasma of DS individuals. Extracellular tau is molecularly complex [
20‐
22], and different assays detect distinct populations of tau alloforms, complicating comparisons of results obtained using different assays. Here, we utilized our in-house NT1 tau assay which we have previously shown to be capable of detecting forms of tau that are significantly elevated in plasma of patients with AD-MCI and mild AD [
21]. Like Aβ42, NT1 detected tau was highest at early age (0–10 years). NT1 levels fell between 11 and 30 years but thereafter increased. This pattern is a mirror image of the Aβ42 results, with both exhibiting pivotal changes between 20 and 40 years.
NfL, a now widely validated marker of neurodegeneration [
18], was relatively low in early life, but in both controls and DS, NfL increased steadily after age 30. Importantly, NfL levels in plasma of individuals with DS started to diverge from control levels in the 31–40-year age group and were most different in the two oldest age groups.
Collectively, our findings demonstrate that plasma measures of amyloid, tau, and neurodegeneration change with age and that the relative differences in these markers are greatest in the 31–40-, 41–50-, and over 50-year age groups. In our study cohort, elevated concentrations of NfL and NT1 measured tau in older individuals with DS are consistent with the following: (i) recent cross-sectional studies that found increased plasma NfL and tau in prodromal and AD dementia in people with DS [
32,
47], and (ii) an increasing prevalence of AD in older individuals with DS [
4,
5]. Notwithstanding the identification of these important trends, our results indicate that it will be impractical to use a single time point measurement of these biomarkers to diagnose AD in DS. Rather, our data support longitudinal assessment of these markers to further evaluate their potential to predict onset of disease.
A particular strength of this study is the use of a relatively large number of DS individuals (n = 100) and age- and sex-matched controls (n = 100) with a broad age range (3 months to 68 years). Another strength is the use of analytically validated methods and testing conditions. The major weaknesses of our study include the fact that the study is cross-sectional and not longitudinal, the use of controls from a biobank, and that we did not include cognitive assessments. Future studies should collect clinical information, such as cognitive measures, APOE status, and concomitant medication, and it may be useful to measure other alloforms of Aβ so as to calculate ratios of different Aβ species (e.g., Aβ42/40, Aβ42/38).
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.