Abstract
Background:
CSF Alzheimer’s disease (AD) biomarkers allow classifying individuals based on their levels of amyloid and neurodegeneration pathologies.
Objective:
To investigate the distribution of AD biomarker profiles from patients suffering from cognitive disorders.
Methods:
We analyzed 3001 patients with cognitive disorders and referred by 18 French memory clinics located in and around Paris. Patients were classified as normal, amyloidosis (A+/N–), amyloidosis and neurodegeneration (A+/N+) or suspected non-AD pathophysiology (SNAP), according to their CSF levels of biomarkers. Analysis were performed for the overall population and stratified by gender, age quintiles, and Mini-Mental State Examination (MMSE) score quintiles. Results were compared to previous findings in cohorts of healthy elderly adults.
Results:
37% of the sample were classified as A+/N+, 22% were classified A+/N–, and 15% as SNAP. The A+/N+ profile was associated with female gender, advanced age, and lower MMSE score, while the A+/N–profile was observed more frequently in men and the distribution was stable across age and MMSE. The SNAP profile showed no association with gender or age, was less frequent in patients with lower MMSE, and had a lower repartition than the one previously reported in asymptomatic populations.
Conclusions:
While A+/N+ patients had the clinical characteristics typically observed in AD, A+/N–patients had a different epidemiological pattern (higher frequency in men, no association with advanced age or lower MMSE). The SNAP profile was less frequent than previously reported in the general elderly population, suggesting that this profile is not a frequent cause of memory impairment in this population.
INTRODUCTION
Alzheimer’s disease (AD) is the most common cause of dementia [1] and is characterized by lesions including extracellular amyloid-β peptide (Aβ) accumulations, neurofibrillary tangles composed of intraneuronal abnormally phosphorylated tau, and neuronal and synaptic losses [2]. The development of AD biomarkers [3] has led to new opportunities for linking cognitive symptoms to AD brain lesions and for detailed study of the presymptomatic phase of the disease [4–8]. 2011 NIA-AA criteria proposed to distinguish between amyloid and neurodegeneration biomarkers [9], and several sub-stages within the pre-symptomatic stage were identified using these biomarkers. Stage 1 was defined by amyloid positive biomarkers only, while stages 2/3 were defined by both amyloid and neurodegeneration positive biomarkers, without or with clinical symptoms, respectively [10]. Suspected non-AD pathophysiology (SNAP) was developed in parallel, for patients with positive markers of neurodegeneration and without evidence of amyloid deposition [11], and an unbiased classification of patients has been proposed more recently, based on amyloid, tau, and neuronal injury biomarkers [12].
AD biomarker profiles are primarily studied in cognitively healthy populations [13–17], as they have been proposed to classify asymptomatic subjects and predict their risk of future cognitive decline [18–21]. Extending this analysis to a population from memory clinics could clarify the meaning of these profiles in patients with complaints and reveal their occurrence in specific pathological conditions commonly addressed in such centers. Here we investigated the distribution of cerebrospinal fluid (CSF) AD biomarker profiles in patients explored for cognitive symptoms in a large and multicentric cohort from nearly 20 memory clinics and compare them to the previously reported distribution in asymptomatic populations.
METHODS
Subjects
CSF analyses are commonly used for routine testing to explore patients with cognitive signs in French memory centers, according to national recommendations [22, 23]. CSF biomarker analyses from 18 hospitals in and around Paris, France, were aggregated in one biochemistry department (Lariboisiere hospital). In this study, we retrospectively included results from all patients for whom a CSF AD biomarker assessment was performed in order to investigate cognitive symptoms from January 1, 2008 to June 30, 2017. The physician was asked to complete a form with basic information, including the suspected clinical diagnosis, the level of education, and the most recent MMSE score for all CSF biomarker evaluations. Diagnoses used for the analysis are therefore the initial supposed diagnoses, before the results of CSF biomarkers, and are based on the initial evaluation of the physicians.
The study was approved by the Ethical Committee of Paris University Hospitals; all participants provided written informed consent.
CSF analysis
Lumbar punctures were performed on fasting patients, typically between 9 and 12 am. All centers used the same model of 10 mL polypropylene tube to collect CSF (January 2008 to November 2012: CML model TC10PCS; December 2012 to June 2017: Sarstedt catalog no. 62.610.201). CSF samples were centrifuged at 1,000 g for 10 min at 4°C within 4 h of collection, and then aliquoted in 0.5 mL polypropylene tubes and stored at –80°C for further analysis. CSF levels of Aβ42, total tau, and ptau-181 were measured using the commercially available sandwich ELISA INNOTEST® sandwich, according to the manufacturer’s procedures (Fujirebio Europe NV, formerly Innogenetics NV). The following cut-offs for pathologic biomarker levels were used: Aβ42 730 pg/mL, total tau 340 pg/mL, and ptau-181 59 pg/mL. These were the optimal cut points (e.g., using the Youden index) to differentiate clinically diagnosed AD patients from other disorders and cognitively normal controls in earlier studies from our group [24, 25]. Concentrations of total tau greater than 1200 pg/mL were well above the detection limit and thus were not recalculated after dilution due to constraints in the procedure. The Alzheimer’s Association quality control program for CSF biomarkers validated the quality of CSF evaluations [26].
CSF biomarker profiles
CSF biomarker profiles were determined based on the NIA-AA staging framework [10]. The amyloidosis profile (A+/N–) was characterized by a low level of CSF Aβ42 and normal levels of CSF tau and CSF ptau 181. The amyloidosis + neurodegeneration profile (A+/N+) was defined by a low level of CSF Aβ42 and increased level of CSF tau and/or CSF ptau 181. Subjects with an increase of CSF tau and/or CSF ptau 181 but normal Aβ42 were classified as SNAP. Those with normal levels of all three biomarkers were classified as “normal”. We further categorized CSF findings based on a recently proposed unbiased classification method, the A/T/N classification [12]. The ‘A’ for Aβ42 biomarker was based on CSF Aβ42 levels, the ‘T’ for tau pathology comes from the CSF ptau 181 level, and the ‘N’ for neurodegeneration is based on CSF levels of tau. Finally, we categorized patients following the “PLM scale”, based on the total number of abnormal CSF biomarkers (0 to 3) [27].
Abnormal values of CSF biomarkers were defined according to local cut-offs (CSF tau and ptau181 above cut-offs and CSF Aβ42 below cut-off). All biomarkers were measured in a single laboratory to prevent inter-site variability [25].
Evidence before this study
We identified cohort studies in which the repartitions of A+/N–, A+/N+, and SNAP has been reported by searching PubMed and Google Scholar with the terms “preclinical AD criteria”, “neurodegeneration”, “SNAP”, and “amyloidosis”, published between January 1, 2011 and July 31, 2017 to compare the biomarker distributions in our population of patients with those previously reported in healthy populations. We then pooled the results of these studies to determine the mean biomarker profile distribution in the healthy elderly population.
Statistical analysis
Proportions were calculated for categorical variables, while means and standard deviations were computed for continuous variables. Statistical significance was assessed using a χ2 test or Student t-test as appropriate.
CSF biomarker profile frequencies were determined overall, by gender, and according to quintiles of age and MMSE distributions in the study population. The relationships between proportion of CSF biomarkers profiles and quintiles of age and MMSE were determined using Cochran-Armitage Trend Test.
Proportions of CSF biomarker profiles observed in our study were compared to those reported previously using a χ2 test.
All resulting p-values were two-tailed and p≤0.05 was considered statistically significant. Statistical analyses were performed using SAS version 9.3 (SAS Institute, Cary, NC, USA).
RESULTS
3,001 patients had an assessment of CSF AD biomarkers in one of the 18 included memory clinics between January 1, 2008 and June 30, 2017. Their characteristics are summarized in Table 1. The mean (SD) age was 69.5 (10.4) years, 50% were women, and the mean (SD) MMSE at the time of the lumbar puncture was 21.7 (5.9). The diagnoses proposed by clinicians requesting lumbar puncture are summarized in Supplementary Table 1. 70% of the patients received a diagnosis of AD (56%) or mild cognitive impairment (MCI) due to AD (14%). Vascular dementia was suspected in 9% of patients, and other etiologies included suspected frontotemporal dementia (5%), normal pressure hydrocephalus (5%), cerebral amyloid angiopathy (4%), or Lewy body disease (3%).
Population characteristics
CSF, cerebrospinal fluid; MMSE, Mini-Mental State Examination; SD, standard deviation. aAPOE genotype determined in a sub-sample of 903 patients.
The repartition of CSF biomarkers profiles in the study population is shown in Fig. 1. 26% of patients had a normal profile, 22% were classified A+/N–, and 37% were classified A+/N+. The SNAP profile accounted for the remaining 15% of patients. A higher proportion of women (42%) than men (32%, p < 0.001) were classified A+/N+. In contrast, the proportion of A+/N–profile was higher in male patients (24%) than in females (19%, p = 0.002). There was no difference in the percentage of SNAP observed in men (16%) and in women (14%, p = 0.14). Patient characteristics according to biomarker profiles are shown in Table 2. Patients with A+/N+ profiles tended to be older (mean: 71.9 years) than those with A+/N– (69.7 years, p < 0.001) or SNAP (70.3 years, p < 0.001) profiles. A+/N+ profile tended to have more female, have a lower mean MMSE scores, and higher frequencies of the APOE ɛ4 allele relative to the other profiles.
Repartition of CSF biomarker profiles in the population.
Characteristics according to CSF Biomarker Profiles
AD, Alzheimer’s disease; CSF, cerebrospinal fluid; MCI, mild cognitive impairment: MMSE, Mini-Mental State Examination; SCI, subjective cognitive impairment; SD, standard deviation. aAPOE genotype determined in a sub-sample of 903 patients. bData was missing for 447 (15%) of the patients.
Table 3 summarizes studies reporting the repartition of biomarker profiles in healthy elderly subjects, for comparison to our population. The proportion of A+N– (x1.6, p < 0.001) and A+N+ (x2.8, p < 0.001) was overrepresented in our patient population compared to asymptomatic subjects, while the proportion of normal (–47%, p < 0.001) and SNAP (–34%, p < 0.001) profiles was lower.
Biomarker profiles in the study population compared to healthy elderly populations
Figure 2 presents the distributions of CSF biomarker profiles according to age and MMSE quintiles. The proportion of A+/N+ increased linearly with age, ranging from 22.1% (first age quintile) to 46.3% (5th quintile, p for trend <0.001). In contrast, the proportions of A+/N– and SNAP were stable across age quintiles. Similarly, A+/N+ profiles were more common in patients with lower MMS scores (p for trend <0.001), whereas no comparable relationship was observed for A+/N– patients (p for trend = 0.37). Finally, SNAP profiles tended to be more frequent in patients with higher MMSE scores (p for trend = 0.01).
Repartition of CSF biomarker profiles according to age and MMSE score.
Supplementary Figure 1 shows the A/T/N and PLM classifications applied to the study population. 26% of the patients had normal levels of all three biomarkers (A–/T–/N–, PLM = 0), while 29% were positive for all three (A+/T+/N+, PLM = 3). 45% of the samples were positive for one (28%, of which 75% were A+/T–/N–) or two (17%) biomarkers. Elevated CSF ptau-181 with normal CSF tau (T+/N–) profiles were uncommon and accounted for only for 2% of the samples.
DISCUSSION
We report the distribution of CSF AD biomarker profiles in this multicentric cohort of 3,001 patients explored for cognitive disorders, and make the following observations:
37% have positive CSF biomarkers for both amyloidosis and neurodegeneration. The majority of these patients were women, and frequency linearly increased with age and with lower MMSE score.
22% were positive for amyloidosis only. The majority of these patients were men and their frequency had no relationship with age or MMSE score.
15% were classified as SNAP. Their frequency decreased with lower MMSE score and none association was observed with gender or age.
Several studies have reported the proportion of AD biomarker profiles in healthy asymptomatic populations [13–17], and few studies on AD patients [28]. As expected, we found that A+/N– and A+/N+ profiles were overrepresented in our population as compared to those reported in studies of asymptomatic populations. The A+/N+ association profile corresponds to our current understanding of AD epidemiology [29]. In contrast, the A+/N– association profile suggests a different epidemiological pattern than that observed in AD. Previous studies have shown that lower CSF Aβ42 levels are commonly associated with other cognitive diseases [8, 30], including vascular dementia or dementia with Lewy bodies, which occur more frequently in men [31, 32]. We therefore hypothesize that the A+/N– profile represents a heterogeneous group of patients that includes AD as well as other cognitive disorders.
We found that the frequency of patients with the A+/N– profile remains stable across patient age, while the A+/N+ profile drastically increases with age. These results are consistent with a previous study of a large cohort of cognitively normal subjects [33]. The authors explained the stable frequency of A+/N– observed in that study by an equilibrium in the rate of subjects entering as A–/N– and leaving as A+/N+ profile. The same explanation could be proposed for the stability across MMSE score quintiles that we observed for A+/N– subjects.
The biomarker-based concept of suspected non-Alzheimer’s disease pathophysiology (SNAP) is recent and remains controversial [12]. Several studies have shown an association between SNAP and cognitive decline [34–37], and the risk may be even higher for SNAP individuals than for A+/N– subjects [12]. However, another recent study found no differences between SNAP and A–/N– subjects in cognitive decline, while the A+/N– and A+/N+ groups declined faster [13]. In this study, we found that the frequencies of both SNAP and normal profiles were lower than those reported in the asymptomatic population, and the percentages further decreased in more severe patients. Our results could argue that SNAP is not a common cause of cognitive disorders, at least in subjects referred to memory clinics, in contrast to the frequency of amyloidosis pathologies in patients with cognitive diseases.
Whether CSF tau is a specific marker of neurodegeneration remains currently discussed. Several studies have also found that tau proteins may be actively secreted from AD-affected neurons in the absence of neurodegeneration, potentially in response to Aβ exposure, resulting in increased CSF concentration [38, 39]. More recently reported markers, like neurofilament light chain, might better fulfil the characteristics of a neurodegeneration marker [40]. As a decreasing in CSF Aβ42 levels may also be observed in other non-AD diseases, the use of Aβ42/40 ratio has previously shown to have some interest [24].
This study has several strengths, including its large size, the inclusion of 18 memory clinics, and the fact that the analyses of CSF were performed in a single biochemistry department. One of the main limitations is that only patients who needed to have a lumbar puncture performed as part of the diagnostic procedure were included, by design. This constraint may prevent generalizing our results to the broader population of patients suffering from cognitive disorders. Future replication of these analyses in different countries and contexts may be useful to better understand the meaning of these profiles in pathological conditions.
Conclusion
In conclusion, this report describes the distribution of CSF AD biomarker profiles in a large and multicentric population of patients with cognitive symptoms, and a comparison with previously reported distributions in healthy elderly populations. We found that patients with only amyloidosis and those with both amyloidosis and neurodegeneration were overrepresented and followed different epidemiological patterns, while the SNAP profile was not common in this population.
DISCLOSURE STATEMENT
Authors’ disclosures available online (&mychar;https://www.j-alz.com/manuscript-disclosures/18-0240r1).
