Age-Dependency of Total Tau in the Cerebrospinal Fluid Is Corrected by Amyloid-β 1–40: A Correlational Study in Healthy Adults

INTRODUCTION

Tau proteins in the human central nervous system (CNS) occur in six isoforms produced from a single gene through alternative mRNA splicing. These proteins play a crucial role in maintaining the stability of microtubules in axons and are established biomarkers for several neurological and neuropsychiatric conditions [1]. While tau phosphorylated at threonine 181 (p-Tau) is considered a specific biomarker of Alzheimer’s disease (AD), total-tau (t-Tau) is interpreted as a non-specific marker of neuronal and axonal damage [2]. t-Tau concentrations in the cerebrospinal fluid (CSF) are elevated in numerous neurological and psychiatric conditions, ranging from acute brain injury like stroke [3] or traumatic brain injury [4] to neurodegenerative diseases like AD [5] and corticobasal degeneration [6]. Most markedly increased levels of t-Tau are observed in rapidly progressing conditions like Creutzfeldt-Jakob Disease [7].

Accordingly, while elevated t-Tau concentrations in the CSF are thought to reflect neuroaxonal damage, it could be assumed that age-related neurotoxic actions like glial activation, mitochondrial dysfunction, and oxidative stress [8] lead to increased levels of t-Tau in the CSF. However, as data is scarce on this, in clinical practice the same cut-off value is used independently of age.

Few studies explicitly address the relationship between age and t-Tau in the CSF of healthy adults and those that do reported divergent results. Sjögren et al. [9] were the first to report a positive correlation in a sample of clinically healthy adults. Participants were required to score above 27 on Mini-Mental State Examination (MMSE) [10]. Similarly, in later studies a positive association between t-Tau and age was found [11, 12]. In contrast, another study found no correlation [13]. Note that in two of the studies [12, 13], no cognitive screening was performed, and inclusion was solely based on patient history and clinical impression. The Mattsson et al. [11] sample was part of a larger study comparing patients with AD, patients with mild cognitive impairment (MCI) and healthy controls; the latter were included if showing no signs of active neurological or psychiatric disease and a MMSE score above 25. Importantly, none of the above studies included additional data on CSF amyloid markers of AD to further rule out underlying AD pathology.

In conclusion, although most studies reported a positive correlation between age and t-Tau in the CSF of healthy adults, the quality of inclusion criteria is questionable and limits the generalizability of the findings.

Besides age-related changes in brain metabolism, other mechanisms could elevate t-Tau levels in the CSF of elderly subjects. Age-associated increased drainage from the cerebrum into the CSF might offer an alternate explanation, which would then result in altered levels of other peptides in the CSF as well. Drainage mechanisms of soluble peptides from the brain interstitial system into the CSF remain poorly understood [14], although recent studies suggest a central role of perivascular pathways called the glymphatic system [15]. The functionality of these mechanisms is suspected to be impaired in the aging brain [16, 17], thereby potentially leading to altered peptide levels in the CSF.

Even more than tau proteins, the role of amyloid-β (Aβ) peptides has been studied extensively in the context of AD and neurodegeneration [18]. Aβ peptides derive from the membrane-bound amyloid-β protein precursor (AβPP) and, when deposited in extracellular plaques, constitute a pathological hallmark of AD [19]. Cleaving of AβPP results in various Aβ-species of different length. Those consisting of 40 amino acids (Aβ₄₀) are the most abundant, followed by those consisting of 42 amino acids (Aβ₄₂). While accumulation of Aβ₄₂ is widely recognized as a key driver of AD pathophysiology [18], Aβ₄₀ has been traditionally interpreted as non-pathogenic for AD [20]. Contrary to Aβ₄₂, Aβ₄₀ CSF concentrations only marginally differ between patients with AD dementia and controls [5]. Hence, the Aβ_{1–42/1 –40}-ratio is widely used in clinical decision making and research to minimize pre-analytical Aβ₄₂ biases and account for between-patient amyloid variations [21]. In view of the above, Aβ₄₀ could be interpreted as a general marker of drainage from the brain’s interstitial system into the CSF, whereas this would not be the case for more specific biomarkers like Aβ₄₂ and tau proteins.

In this study, we aimed to provide further insight into the relationship between age and t-Tau in the CSF by using Aβ₄₀ as a proxy for drainage mechanisms of peptides into the CSF. Specifically, we explored relationships of age, t-Tau, and the ratio of t-Tau and Aβ₄₀ to disentangle putative age-related from putative drainage-related effects. To ensure the exclusion of patients with neurological or psychiatric conditions, we employed rigorous inclusion criteria for study participation and thereby present data from a unique and carefully selected sample of cognitively healthy adults.

MATERIALS AND METHODS

The study was conducted in accordance with the Declaration of Helsinki [22] and approved by the ethics committee of the Technical University of Munich (TUM), Munich, Germany (project code: 29/17 S). All study participants provided written informed consent.

RESULTS

Correlation analyses

Correlation analyses revealed a positive association of t-Tau and age (rho = 0.433, p < 0.001, significant after Bonferroni-correcting alpha-levels in the core sample; 0.05/4 = 0.0125) and age was similarly correlated with Aβ₄₀ (rho = 0.581, p < 0.001). In contrast, the t-Tau/Aβ₄₀ ratio was not correlated with age (rho = –0.026, p = 0.825), resulting in a significant difference between the correlation coefficients of t-Tau and age and the t-Tau/Aβ₄₀ ratio and age (p < 0.001). Corresponding scatterplots are depicted in Fig. 1. Furthermore, when controlling for Aβ₄₀ in a partial correlation analysis of t-Tau and age, no significant association was observed (rho = 0.190, p = 0.103).

OPEN IN VIEWER

Fig. 1

Scatterplots and correlation coefficients for age and t-Tau as well as age and t-Tau/Aβ₄₀ ratio in the core sample (n = 76, normal Aβ_42/40 ratio).

OPEN IN VIEWER

Fig. 2

Secondary analyses: Scatterplots and correlation coefficients for age and t-Tau as well as age and t-Tau/Aβ₄₀ ratio (A: n = 82 for participants with normal Aβ₄₂ concentrations, B: n = 36 for participants who underwent extensive neuropsychological evaluation with normal results and had normal Aβ_42/40 ratios).

Secondary analyses revealed similar coefficients when using normal Aβ₄₂ concentrations for inclusion regarding t-Tau and age (rho = 0.457, p < 0.001, significant after Bonferroni-correcting alpha-levels in this subsample; 0.05/3 = 0.0167) and the t-Tau/Aβ₄₀ ratio and age (rho = –0.019, p = 0.886). A partial correlation analysis of t-Tau and age controlled for Aβ₄₀ showed a non-significant association (rho = 0.218, p = 0.051) for this subset of patients. Comparable results were obtained when examining the subsample of neuropsychologically validated cognitive healthy subjects for t-Tau and age (rho = 0.405, p = 0.010, significant after Bonferroni-correcting alpha-levels in this subsample; 0.05/3 = 0.0167) and for the t-Tau/Aβ₄₀ ratio and age (rho = 0.083, p = 0.631), as well as in a partial correlation analysis of t-Tau and age controlled for Aβ₄₀ (rho = 0.151 p = 0.386). Testing for differences of correlation coefficients between primary and secondary analyses consistently revealed non-significant results for differences in correlations between t-Tau and age, as well as for differences in correlations between the t-Tau/ Aβ₄₀ ratio and age.

Note that in all subsamples age was correlated with Aβ₄₀ (rho = 0.439 in the core sample, rho = 0.403 for participants with normal Aβ₄₂ concentrations and rho = 0.429 for participants who underwent comprehensive neuropsychological testing, all ps < 0.01).

DISCUSSION

In this study, we found a positive correlation between t-Tau and age in a sample of carefully selected healthy adults ranging from 30 to 85 years of age, which is in line with previous work. Crucially, this association was no longer present when correcting t-Tau for Aβ₄₀ concentrations or controlling the correlation of t-Tau and age for Aβ₄₀. These findings were replicated applying varied inclusion criteria regarding amyloid biomarkers for AD in order to exclude underlying asymptomatic AD and applying rigorous criteria for extensive cognitive testing in order to rule out subtle cognitive deficits.

Our results question the validity of the reported association of age and t-Tau in the CSF of normally aging subjects [9, 12] and raise issues regarding the role of Aβ₄₀ in the context of interpreting CSF biomarkers associated with AD and other neurodegenerative diseases. While Aβ₄₀ has traditionally been used to correct pre-analytical Aβ₄₂ biases and to account for interindividual amyloid variations [21], it might have additional diagnostic value in clinical decision making and research settings. Since clinicians are constantly presented with ambiguous constellations of CSF biomarkers when testing patients suspected of suffering from neurodegenerative diseases, there is a clear need for better understanding factors influencing CSF peptide levels. While drainage-mechanisms from the brain’s interstitial system into the CSF remain a matter of ongoing debate [14], considering Aβ₄₀ as a possible proxy for drainage-related interindividual variance in peptide concentrations could be of heuristic value. Note, however, that despite known diurnal fluctuations of amyloid peptide in CSF and plasma [31], our study does not provide direct in vivo evidence for Aβ₄₀ as a marker of peptide drainage into the CSF.

In line with this, the diagnostic utility of the t-Tau/Aβ₄₀ ratio should be explored in AD and other neurodegenerative diseases. For example, the diagnostic accuracy of CSF-t-Tau in distinguishing patients with AD at an asymptomatic or MCI stage from healthy aging might be underestimated when not taking drainage-related interindividual differences into account.

In contrast to previous studies examining t-Tau in the CSF of healthy subjects, we excluded subjects with abnormal amyloid profiles of CSF biomarkers for AD (as the most common underlying cause of dementia; [32]). Furthermore, we ruled out other infectious/inflammatory CNS processes by analyzing samples regarding cell count and lactate levels. In order to minimize influences of undetected psychopathology on the results of cognitive testing [33], we also included measures of affective symptoms and excluded depression. We strongly recommend following strict inclusion and exclusion criteria when examining biomarkers of neurodegeneration in healthy subjects to maximize validity of results.

Our study has some limitations, which should be addressed in further research. As our sample size is moderate and patients were recruited at a single hospital, results should be replicated with a larger number of patients in a multi-center approach. Additionally, the drainage mechanisms of other, non-amyloid proteins should be evaluated in further studies as Aβ₄₀ cannot be viewed as completely neutral to AD pathology. For instance, Aβ₄₀ has been studied as a CSF marker for cerebral amyloid angiopathy [34]. As analysis of AD-related peptides from more easily accessible body fluids than CSF is gaining attention [35], the relationship between age and individual biomarkers of neurodegeneration in blood and saliva should be examined in future studies.

In this study, we found a correlation of CSF t-Tau and age in cognitively healthy adults, which was no longer present when correcting t-Tau for Aβ₄₀ concentrations. The results call into question the reported association of age and t-Tau. Furthermore, they suggest diagnostic utility of Aβ₄₀ as a proxy for CSF drainage mechanisms when interpreting biomarker concentrations of neurodegeneration in research and clinical settings.