Correlation between cognition and plasma noradrenaline level in Alzheimer’s disease: a potential new blood marker of disease evolution

Recent evidence showing degeneration of the noradrenergic system in the locus coeruleus (LC) in Alzheimer’s disease (AD) has motivated great interest in noradrenaline (NA) as a potential brain hallmark of the disease. Despite the current exploration of blood markers for AD, the deregulation of the plasma NA concentration ([NA]plasma) in AD is currently not well understood. This retrospective study includes a cohort of 71 patients (32 AD patients, 22 with other dementia and 17 without dementia) who were given consultations for memory complaints in the Cognitive Neurology Center of Lariboisière (Paris) between 2009 and 2014. As previously described in brain tissue, we show for the first time a linear correlation between [NA]plasma and Mini Mental State Examination (MMSE) score in AD patients. We observed that high [NA]plasma in AD patients was associated with higher [Aβ1–42]CSF than in other AD patients with [NA]plasma similar to NC patients. In parallel, we observed a lower (p-Tau/Tau)CSF in AD patients with low [NA]plasma than in non-AD patients with [NA]plasma similar to [NA]plasma in NC patients. Our data suggest that [NA]plasma could be a potential biomarker of disease evolution in the context of AD and could possibly improve early diagnosis.


Introduction
Alzheimer's disease (AD) is the most common cause of dementia among adults over 65 years old and is characterized by progressive impairment in cognitive function and behavior 1 . The neuropathological hallmarks of the disease include amyloid plaques, formed by extracellular accumulation of amyloid β peptide (Aβ  ) and neurofibrillary tangles, formed by intraneuronal aggregates of phosphorylated Tau protein p-181 (p-Tau) 2 . Cerebrospinal fluid (CSF) core biomarkers (Aβ 1-42 , p-Tau, and total-Tau), as well as amyloid and Tau positron-emission tomography, reliably reflect AD neuropathological brain lesions 3,4 , leading to their inclusion in research diagnostic criteria. However, acquiring these measurements is costly and invasive. In addition, CSF biomarkers are sometimes not sufficient to differentiate AD from other types of dementia, such as dementia with Lewy bodies 5 . Indeed, differential diagnosis remains a major challenge in clinical practice 6 . Thus, addressing the need to develop novel and inexpensive diagnostic techniques has become a focal point of Alzheimer's research 2 .

The noradrenergic system and AD
Many studies have shown that the abnormal structure and function of noradrenergic neurons are closely related to AD pathophysiology [6][7][8] . The noradrenergic system plays pivotal regulatory roles in various behaviors, including selective attention, memory storage and retrieval, vigilance, and mood 9 . Most of the noradrenergic neurons in the CNS originate from the locus coeruleus (LC) and project to different areas of the brain, such as the cortex, hippocampus, amygdala, thalamus, and hypothalamus 10 . Hyperphosphorylation of Tau protein in the LC appears at an early stage of AD pathogenesis 11 , and the number of neurons in the LC has been shown to progressively decrease during the disease, beginning at the prodromal stage of AD 8,[12][13][14][15][16][17] . Based on the current literature, one of the ideas that has emerged is that the onset of AD is preceded by abnormal hyperactivation of the LC, resulting in oversecretion of NA in the cortex 18 . The hyperactivation of noradrenergic receptors in the cortex leads to a cortical accumulation of Aβ 1-42 plaques 19 . Following this accumulation, the dendritic spine density of noradrenergic neurons and noradrenaline (NA) secretion begin to decrease due to lack of stimulation and/or (glial) inflammatory processes 18 . However, this working hypothesis requires further research in this area and supporting evidence. NA is a catecholamine that acts as a neuromodulator when synthesized by noradrenergic neurons and as a hormone when synthesized by the sympathetic system. NA deregulation in AD was first identified at the end of the 1970s by Adolfsson et al., who showed that the NA concentration in postmortem brain tissue was lower in AD patients of~75 years of age than in age-matched controls and correlated with dementia score 20 . Interestingly, a recent publication reported a correlation between the NA concentration in a specific cortical brain region of AD patients and MMSE score 21 . In addition, NA CSF levels in advanced AD patients differ from those in patients with mild to moderate severity and those in healthy older subjects 22,23 . These observations imply that cognitive decline could be timely correlated with coeruleo-cortical network deregulation and disconnection 18 . The early increase in neuronal NA secretion before the onset of AD that is later followed by noradrenergic neuron degeneration could be related to disease evolution 11,24 . At the peripheral level, NA circulates through the bloodstream and is known to be involved in the body response to acute stress by modulating, among other physiological parameters, heart rate, blood pressure, and muscle contraction 25 . Interestingly, the LC is known to regulate sympathetic activity 26 by modulating postsympathetic noradrenergic neurons targeting the heart and blood vessels 27

Early stage of AD: CSF and plasma biomarkers
Investigations of potential treatments for AD have revealed the need to define and characterize an early stage of the disease, where therapeutic treatments could be the most effective. Recent studies have attempted to identify new biomarkers that can identify AD patients before the onset of advanced symptoms. The term mild cognitive impairment (MCI) was used for the first time in the 1980s to define patients at an intermediate stage between normal aging patients and patients with dementia 31 . Recently, the National Institute on Aging and Alzheimer's Association (NIA-AA) specified the state "MCI due to AD" to clearly identify a very early stage of this disease 32 . Among other clinical criteria 33 , CSF biomarkers are required in the determination of MCI due to AD, such as CSF Aβ 1-42 , total-Tau, and/or p-Tau concentrations 32 . Several studies in the literature have shown a significant difference in CSF Aβ 1-42 34-36 , total-Tau [35][36][37][38][39] , and p-Tau 37 concentrations between MCI and AD patients, describing a decrease in the CSF Aβ 1-42 concentration ([Aβ  ] CSF ) and an increase in the CSF total-Tau and p-Tau concentrations ([Tau] CSF and [p-Tau] CSF ) in AD patients. However, these differences are not always statistically significant for CSF Aβ 1-42 37,40 , total-Tau 34,40 , and p-Tau 34,35,39,40 . A higher CSF p-Tau/total Tau ratio ((p-Tau/Tau) CSF ) (or lower (Tau/p-Tau) CSF ) was also previously described in AD patients in comparison with healthy controls 41,42 . Furthermore, (p-Tau/Tau) CSF has been shown to be lower in AD patients than in control patients and patients with fronto-temporal dementia 43 . However, no difference in (p-Tau/Tau) CSF was found between patients with mild AD and those with moderate AD 41 . (p-Tau/Tau) CSF was also associated with the rate of cognitive decline, showing a negative dose-dependent relation between (p-Tau/Tau) CSF and yearly MMSE score fold change during AD 44 . In other words, a low (p-Tau/ Tau) CSF in patients with AD is associated with a faster cognitive decline than that observed in other AD patients with higher (p-Tau/Tau) CSF . This association implies that AD patients with low [Aβ  ] CSF and elevated [Tau] CSF and [p-Tau] CSF are more likely to be in an advanced stage of the disease.
Although these results are conflicting, the CSF AD biomarker profile seems helpful in the differentiation of MCI patients from AD patients and provides information about whether the patients is in an early or late stage of the disease.
The need to find alternative body fluids for biomarker identification in AD has recently led investigators to analyze currently known CSF biomarkers for AD in plasma [45][46][47][48][49][50][51][52][53][54][55][56][57] . Studies in this area have shown conflicting results with (i) no difference between AD and control patients 47,49,50 , (ii) differences in plasma Aβ 1-42 45,58 , Tau 57 and p-Tau 54 concentrations between AD and control patients, and (iii) differences in plasma Aβ 1-42 53,58 , Tau, and p-Tau 54 levels among patients with MCI due to AD, patients with AD, and control subjects. Other articles have reported that the plasma Aβ 1-42 concentration might change during disease evolution, showing abnormally high concentrations at the preclinical stage or beginning of cognitive decline that decrease progressively during AD progression 51,55 . Moreover, other studies have shown different linear correlations in the context of AD (i) between imaging data of amyloid plaque deposition in the brain and CSF 59 or plasma 46,48 biomarker concentrations and (ii) between CSF and plasma biomarker concentrations 45,47,50,57,58 . The mechanisms behind brain AD biomarker clearance pathways (blood-brain, CSF-brain and blood-CSF barriers) are not yet understood, and it is not clear if those barriers are altered during the disease. Nonetheless, some articles suggest that biomarkers in plasma could mirror brain and CSF biomarker deposition and metabolism 60 . However, some studies did not observe any correlation between CSF and plasma biomarker concentrations 56 . There are a variety of possible factors that could explain the inconsistent conclusions and concentration value heterogeneity among studies. For example, technical (antibody used, time and temperature of sample conservation, hydrophobic and albumin binding properties of Aβ, low concentration in comparison with CSF, etc.) and physiological (effect of age and diet, peripheral origins, kidney and hepatic clearance, etc.) reasons 51,58 may make determining plasma Aβ 1-42 level alterations in the context of AD difficult. For these reasons, the clinical utility of plasmatic Aβ 1-42 and Tau has not yet been demonstrated, and the need to identify other plasma AD biomarkers is crucial for AD diagnosis.
In this retrospective study, we explored [NA] plasma in patients between 58 and 79 years old consulting for memory complaints for the first time. Our study examined the relationship between [NA] plasma and concomitant diagnostic criteria such as MMSE score and CSF biomarker profile (Aβ 1-42 , Tau and p-Tau). Due to the deregulation of noradrenergic transmission in the brain during AD, we wanted to determine whether [NA] plasma could be correlated with clinical parameters reflecting the stage of the disease at the cognitive (MMSE score) and molecular (Aβ 1-42 , Tau, and p-Tau CSF biomarkers) levels.

Study population
All patients presented to the Cognitive Neurology Center of Lariboisière (Paris) for their first consultation between 2009 and 2014. Patients involved in this study were between 58 and 79 years old at the time of blood sampling. MMSE score and lumbar puncture were performed the day of blood sampling or <1 month later. The MMSE is a worldwide commonly used, easy to apply and rapid screening tool lasting <10 min. The MMSE score provides supporting information for dementia diagnosis that evaluates global cognitive impairment for all causes of dementia 61 . Briefly, the paper-based test consists of 11 orally administered questions with verbal and written responses concerning different domains (attention, memory, orientation, language, and ability to follow verbal and written commands) with a maximum score of 30 (normal cognition) and lower scores highlighting a severe deficit 62 . The original article published in 1975 defined 24 as a cutoff score for normal cognition with a sensitivity and specificity of 87% and 82.6%, respectively 61 . In a more recent study, the mean MMSE score of the population older than 90 years without dementia was reported to be 26.6, whereas the cutoff score for normal cognition was reported to be 23.3, very similar to the cutoff score for normal cognition in the younger population (~23 or 24 points) 62,63 . In the context of AD, the MMSE score is used to assess disease severity and is occasionally used to estimate efficacy in clinical drug trials 64 . Previous studies have shown that MMSE scores decrease during AD progression at different rates depending on disease progression and patient education 65 . More generally, patients in the moderate to severe stages of AD have an MMSE score below 20 66 . Sample size was calculated based on results from previous articles highlighting a significant difference in plasma catecholamines between AD and non-AD patients. Raskind et al. 23 observed a significant difference between advanced AD and control patients with a small sample size (<20 patients). Therefore, we based our calculation for sample size on the article of Umageki et al. 28 that examined adrenaline levels. Considering a common standard deviation of 18.46 pg/mL, a sample size of 68 patients is necessary to show a difference of 12.69 pg/mL between the two groups with a power of 80% and an α of 5%. Seventy-one patients were included in this retrospective study: 32 AD patients (diagnoses were performed according to NIA-AA guidelines 4 ), 22 other dementia (OD; frontotemporal dementia, vascular dementia, or dementia with Lewy bodies) patients, and 17 neurological control (NC) patients. NC patients were defined as those with memory complaints, mental depression or anxiety but for whom no dementia was diagnosed. The following cutoff values for core AD CSF biomarkers were used as supportive criteria for dementia due to AD: Aβ1-42 (<550 pg/mL), total-Tau (>400 pg/ mL), and p-Tau (>50 pg/mL). Demographic information, MMSE scores, presence of the APOE ε4 allele, CSF biomarkers (Aβ 1-42 protein, total Tau, and p-Tau protein) and concomitant medications of the studied cohort are shown in Table 1.

Plasma NA quantification
Patients fasted overnight (for~12 h) before blood collection and were in the decubitus position during sampling. Plasma samples were purified and analyzed with a reagent kit for HPLC analysis of catecholamines in plasma (Chromsystems, order #5000) according to the manufacturer's instructions. Briefly, blood samples were stabilized with glutathione, and plasma was isolated less than 1 h after blood sampling by centrifugation. Plasma samples were stored at −80°C. After thawing, 1 mL of plasma was used to extract catecholamines for dosage by high-performance liquid chromatography coupled with electrochemical detection. Experimenters did not know the corresponding group of the sample during dosage.

CSF biomarker quantification
Lumbar punctures were performed on fasting patients, typically between 9 and 12 a.m. CSF samples were centrifuged at 1 g for 10 min at 4°C within 4 h of collection, aliquoted in 0.5-mL polypropylene tubes and stored at -80°C for further analysis. CSF levels of Aβ 1-42 , total Tau, and p-Tau were measured using the commercially available sandwich ELISA INNOTEST ® , according to the manufacturer's procedures (Fujirebio Europe NV, formerly Innogenetics NV).

Statistical analysis
Depending on the normality of the data (D'Agostino-Pearson normality test), the results are presented as the mean with standard deviation (SD) (

Characterization of the study cohort
The studied groups did not significantly differ by sex ratio, age, or concomitant treatments ( Table 1). As expected, they differed by MMSE score and by APOE ε4 carrier status (Table 1). Clinical diagnosis of AD made by the neurologist was based on age, MMSE score, and CSF biomarkers, according to NIA-AA guidelines 4 . AD patients had significantly lower Aβ 1-42 , higher p-Tau, and higher total-Tau CSF concentrations than OD and NC patients (Table 1).
Correlation between plasma NA concentration and cognitive MMSE score in AD patients As previously described in a specific cortical brain region 21 , we observed a significant linear correlation between [NA] plasma at the peripheral level and MMSE score in AD patients (Spearman's correlation, r = 0.4426 (95% CI: 0.1004-0.6912); p value = 0.0112; equation: Y = 126.3*X + 158) (Fig. 1a). However, this correlation was not observed in non-AD patients (Spearman's correlation, r = −0.01385 (95% CI: −0.3365 to 0.3117); p value=0.9333; equation: Y = −44.63*X + 3556) (Fig.  1b). These results suggest a possible link between [NA] plasma and cognitive decline in AD patients. Nonetheless, we observed no difference in [NA] plasma among the AD, OD and NC patients (Table 1). We discriminated patients with an MMSE score above (≥) and below (<) 24, which correspond to the cutoff score for normal cognition in the literature 67 . In non-AD patients, we observed no difference in the distribution of [NA] plasma between patients with an MMSE score above and those with a score below 24 (Kolmogorov-Smirnov test, p value = 0.7459) (Fig. 2a). On the other hand, we found a significant difference between the distribution of [NA] plasma in AD patients with an MMSE score above and those with a score below 24 (Kolmogorov-Smirnov test, p value = 0.0260) (Fig. 2b). Moreover, we observed that the median [NA] plasma of AD patients with an MMSE score above 24 was significantly higher than the median [NA] plasma of non-AD patients with a similar MMSE score (≥24) (Mann-Whitney test, p value = 0.0287) and lower (<24) MMSE score (Mann-Whitney test, p value = 0.0136) and than the median [NA] plasma of other AD patients (Mann-Whitney test, p value = 0.0177). We observed no difference between the median [NA] plasma of AD patients with an MMSE score below 24 and that of non-AD patients with similar (Mann-Whitney test, p value = 0.8757) or higher (Mann-Whitney test, p value = 0.3851) MMSE scores (Fig. 2c). Taken together, these results support the idea that there is a subpopulation of AD patients with an MMSE score above 24 and a higher median {NA] plasma than other patient groups, implying that a mild cognitive decline in AD depicted by MMSE score is associated with an elevated [NA] plasma .

NA plasma level in AD is associated with MMSE score and CSF biomarker profile
To determine whether [NA] plasma reflects early disease evolution, we examined whether there was a correlation between the distance from the NC patient median [NA] plasma and the CSF biomarker profile. Because of the non-normal distribution of [NA] plasma data, we decided to use logarithmic data as previously described 68 (in the following text, all statistical tests for CSF biomarkers were performed on logarithmic data). This distance was defined as the absolute value of ln(NA) plasma -<ln(NA) plasma/N >, with <ln(NA) plasma/N > the median value of logarithmic [NA] plasma from NC patients (<ln(NA) plasma/NC ≥ 7.849). We found a significant positive linear correlation between this absolute distance and [Aβ  ] CSF in AD patients (Pearson's correlation, r = 0.3694 (95% CI 0.01047 to 0.6439); p value = 0.0446; equation: Y = 0.6449*X + 5.695) (Fig. 3a), but this correlation was not observed in non-AD patients (Pearson's correlation, r = −0.1629 (95% CI −0.4706 to 0.1801); p value = 0.3498; equation: Y = −0.1662*X + 6.697) (Fig. 3b). Looking at the relative distance of ln[NA] plasma from <ln(NA) plasma/NC > (negative or positive when respectively lower or higher than the NC patients median), we observed a negative correlation that does not reach significance for negative distance

Discussion
There is a significant amount of literature on deregulation of cerebral, plasma, and urine NA concentrations in murine models of AD [69][70][71][72] and in human AD patients 21,28,29,73,74 . In mouse models, NA has been implicated in the microglial phagocytosis of Aβ 1-42 plaques and cognition 24,75 . In humans, NA levels in cortical brain regions have been correlated with dementia score 20 and MMSE score 21 , which decreases during AD development 67 [34][35][36][37]40 . There have also been conflicting results concerning CSF Tau and p-Tau concentrations, showing no difference or increased levels in late-stage AD compared with MCI [34][35][36][37]40,41,76 . However, the differences observed were not always significant. The absence of statistical significance could be due to several parameters that differ among the different studies, such as the precision of the chemical dosage, sample storage, age of patients, sex ratio of the cohort groups, MMSE score, or advancement of the disease, involving stages of AD more or less distant from MCI. Moreover, the temporal evolution of biomarkers and cognitive markers of AD have previously been shown to differ according to disease stage (asymptomatic, MCI and dementia stage) 77 . We observed a lower (p-Tau/Tau) CSF in AD patients with low [NA] plasma than in non-AD patients with [NA] plasma similar to [NA] plasma in NC patients. A low (p-Tau/Tau) CSF has also been associated with faster cognitive decline 44 , suggesting a possible link between this ratio and disease severity. In this way, low (p-Tau/ Tau) CSF associated with low MMSE score and low [Aβ  ] CSF is more likely to describe an advanced stage of AD pathology.
Regarding MMSE score, [Aβ 1-42 ] CSF , and (p-Tau/ Tau) CSF , it is tempting to speculate that AD patients with high or low [NA] plasma represent an early or late stage of AD, setting [NA] plasma as a potential marker of disease evolution. However, the link between the brain and [NA] plasma alterations during AD is not yet understood. We hypothesize that the [NA] plasma increase observed in AD patients could be due to (i) NA oversecretion in the CNS and disruption of the blood-brain barrier that may occur in AD 78 , (ii) release of cerebral NA in the blood through the CSF interface, with the potential of a linear correlation between [NA] CSF and [NA] plasma 23 , (iii) deregulation of NA secretion into the blood by the sympathetic system regulated by the LC 26 , and (iv) NA oversecretion from sympathetic noradrenergic postganglionic neurons to targeted organs and leakage or spillover into the blood 79 . We did not approach the concept of cognitive reserve in this article. Cognitive reserve is the phenomenon that maintains cognitive function despite advanced physiological symptoms of AD pathology in the brain 80 . Robertson developed the idea that LC-noradrenergic system activity may be the biological mechanism explaining the gap between cognitive function and pathophysiological observations of the disease (especially amyloid plaques) 81 , which has been further supported by several articles [82][83][84] . The pupillary response, a potential marker of NA-LC activity, illustrates that NA activity is enhanced by cognitively challenging stimuli. Knowing that NA seems to mediate neurocompensatory and neuroprotective effects, variables of cognitive reserve (education, enriched environment/ novelty, social interactions, mental activity) may lessen AD pathology through noradrenergic pathway activation 81 . Thus, this pathway represents a good candidate for the missing link in the cognition-brain pathology discrepancy observed in AD. Robertson has highlighted the fact that cognitive reserve phenomena through the action of NA in AD could have a direct effect on AD symptoms (antiinflammatory action, inhibition of amyloid aggregation, etc.) and/or a compensatory effect (increased connectivity, network reorganization). This complex question remains open and needs to be explored with further studies.
In this study, we showed for the first time that [NA] plasma was correlated with MMSE score in the context of AD. We described particular profiles of MMSE score and CSF AD biomarkers for AD patients with high and low [NA] plasma . [NA] plasma deregulation could hypothetically reflect overactivation and regression of the LC during AD. These results imply that (1) [NA] plasma could support the early diagnosis of AD, (2) [NA] plasma could help to better characterize AD patient profiles during disease evolution, and (3) the link between LC degeneration and [NA] plasma in the context of AD needs to be further investigated. Combined with the MMSE score, [NA] plasma could help the patient be referred to a specialist before the appearance of advanced AD symptoms. For example, physicians could advise patients with memory complaints, good MMSE scores and high [NA] plasma to consult a neurologist for further examinations. Thus, this process could favor fast and early diagnosis of AD, which opens new research potentials for blood biomarkers in AD.
Ethics approval and consent to participate Patients gave their informed and written consent to have their samples stored in an officially registered and ethically approved biological collection that has been approved by the Ethics Committee of Paris University Hospitals (CEERB [Comité d'Ethique En Recherche Biomédicale], Bichat University Hospital, Paris, France).

Conflict of interest
Alzohis is a company that has activities related to the submitted work. This study and this publication were produced and written in a responsible and ethical manner.