CSF Tau Protein in Alzheimer’s Disease and other Neurological and Psychiatric Diseases

Research Article

Austin Alzheimers J Parkinsons Dis. 2014;1(2): 10.

CSF Tau Protein in Alzheimer’s Disease and other Neurological and Psychiatric Diseases

Rozenstein-Tsalkovich L1, Kahana E2, Shenker A2, Baitcher F1, Cohen OS3, Kahana-Merhavi S1, Chapman J3, Korczyn AD4, Aharon-Perez J5, Milo R2, Rosenberg A6, Treves T7, Wertman E8, Abramsky O1, Meiner Z1,9 and Rosenmann H1*

1Departments of Neurology, Hadassah Hebrew University Medical Center, Israel

2Department of Neurology, Barzilai Medical Center, Israel

3Department of Neurology, Sheba Medical Center, Israel

4Sieratzky Chair of Neurology, Tel Aviv University, Israel

5Department of Neurology, Rambam Medical Center, Israel

6Department of Neurology, Sorasky Medical Center, Israel

7Department of Neurology, Rabin Medical Center, Israel

8Neurology Services, Clalit HMO, Israel

9Physical Medicine and Rehabilitation, Hadassah Hebrew University Medical Center, Israel

*Corresponding author: Rosenmann H, Departments of Neurology, Hadassah Hebrew University Medical Center, Jerusalem 91120, Israel.

Received: September 08, 2014; Accepted: October 14, 2014; Published: October 15, 2014


Total tau (T-tau) in the Cerebrospinal Fluid (CSF) is among the most reliable markers for the diagnosis of Alzheimer’s disease (AD). However, the high inter-center variation in biomarker concentrations points to the need for setting up specific diagnostic cut-offs for each population. We analyzed the level of T-tau in the CSF of 490 patients affected by AD or other neurological and psychiatric diseases, among the Israeli population. The T-tau levels were significantly higher in the AD group than in the other non-AD diseases, particularly in the other common dementia Fronto Temporal Dementia (FTD), as well as in psychiatric diseases. Receiver-Operating-Characteristic (ROC) analysis provided a ≥240 pg/ml cut-off for discrimination between AD and other non-AD diseases {sensitivity 68%, specificity 60%, negative-PV (predictive-value) 84.4%}, and psychiatric diseases (60%, 80%, positive-PV 93.4%), and from FTD (60%, 61.1%, positive-PV 85.8%), respectively. In spite of low PVs, T-tau levels in the AD patients were also higher than in vascular-dementia, Parkinson’s-disease, epilepsy, with similar trends relative to multiple-system-atrophy, dementia with Lewy-body-disease, autoimmune and other degenerative diseases, while comparable with metabolic and acute neurological diseases. Our previously reported ≥1,000 pg/ml cut-off for Creutzfeldt–Jakob Disease (CJD) diagnosis allows discrimination from AD (sensitivity 75%, specificity 92.7%, PPV 86.56% NPV 85.8%). An inverse correlation was noticed between CSF T-tau levels and Mini-Mental-State-Examination (MMSE) scores in AD and in VD.

Using ≥240 pg/ml as a cut-off we showed here that the T-tau level in CSF could be an indicator for differentiation of AD from psychiatric diseases and from FTD in our population. While also informative at ≥1000 pg/ml for CJD, the T-tau level was less informative for discrimination of AD from other neurological diseases. Combining the T-tau level in the CSF with other parameters (additional CSF markers, as well as genetic and clinical, including imaging parameters) may provide a stronger indication for AD.

Keywords: Neurofibrillary; Olivopontocerebellar; Neurological diseases; Encephalomyopathy


There is accumulating evidence that the major neuropathological features characteristic of the Alzheimer’s disease (AD) brain - are generally reflected in the CSF. A higher amyloid plaque burden in the brain is reflected as lower amyloid beta (Aβ) levels in the CSF [1-4]. A higher CSF tau protein level represents axonal injury and cell death {particularly reflected by T (total)-tau}, as well as a higher neurofibrillary tangle burden {represented by both T- and phosphorylated (P)-tau} [4-6]; all of these markers are considered to be the “core CSF AD markers”. Conflicting results have been reported on the association of the CSF Aβ levels with the cognitive /clinical status [7-11]. Yet, there is more evidence for an inverse association between the tau protein and the cognitive status [7,9,10,12-15].

Many studies have shown the diagnostic value of both Aβ42 and the tau protein in the CSF for discrimination between AD patients and non-demented subjects [16-33]. As for the diagnostic value of CSF markers in discrimination between AD and other dementias, there are conflicting results regarding CSF Aβ42, specifically for differentiation between AD and dementia with Lewy-Body Disease (DLB) and Vascular Dementia (VD) [34,35]; yet, there are more consistent results showing a lower level of the tau protein in the non- AD dementias relative to the AD dementia (yet with overlap) [23,35- 41].

CSF studies in AD patients and non-demented controls from different centers report different biomarker concentrations, reference ranges and diagnostic cut-offs; in some studies, CSF marker levels in AD patients even exceeded (in the case of amyloid) the levels in controls of other studies [42,43]. Several multi-center studies have been conducted {such as the DESCRIPA study by Visser et al. [44], the ADNI study by Shaw et al. [28], the European-ADNI (E-ADNI) by Buerger K et al. [45] and the multi-center study of Mattsson et al. [19]}, showing lower diagnostic accuracies than those of homogenous mono-center studies, presumably related to inter-center variations. The very wide range of cut-offs of the AD CSF markers among centers (such as for tau, ranging from 195 pg/ml to 450 pg/ml [43]), points to the need for setting up specific diagnostic cut-offs to be used in each center, and validate the diagnostic accuracy of the CSF markers in each of the studied populations.

Since tangle pathology is in good correlation with clinical dementia [46-48] and since T-tau in the CSF is associated with tangle-pathology as well as with neuronal loss, the T-tau is considered to be the most commonly used diagnostic tool in many centers. We report here the levels of the T- tau in the CSF of AD patients and of other dementias as well as of patients with other neurological or psychiatric diseases among the Israeli population referred to our laboratory, in the years 1999-2007, which is a national referral center for CSF analysis of neurodegenerative markers.



The study population included a total of 490 patients affected by:

AD: AD (n=124), mixed dementia (AD + vascular dementia (n= 14)

Non-AD dementias: VD (n=114), Fronto Tempotal Dementia (FTD) (n=36) {including tauopathies (Pick’s disease, cortico basal degeneration, progressive supranuclear pulsy (n=14)}, DLB (n=11); Parkinson’s disease with dementia (PDD) (n=6);

Acute neurological diseases: encephalitis (n=13), cerebrovascular accident (CVA) (n=17), intoxication (alcoholism, drug addiction) (n=16), brain tumor (n=5), and paraneoplastic syndromes (n=6);

Other neurological diseases: Degenerative diseases {multiple system atrophy (MSA) (n=14), PD (n=8), amyotrophic lateral sclerosis (ALS) (n=5), Huntington’s disease (HD) (n=2), spinocerebellar ataxia (n=3), cerebellar atrophy (n=2), cerebellar degeneration (n=3), and others (olivopontocerebellar degeneration cerebellar ataxia, pontocerebellar atrophy, cerebellar atrophy, Shy Drager, neuro-axonal sclerotic encephalopaty, CNS degeneration, prog bulbar palsy) (n=8)}, epilepsy (n=16), metabolic diseases {hepatic/pancreatic encephalopathies, mitochondrial encephalomyopathy, lactic acidosis and stroke (MELAS)} (n=14), autoimmune diseases (vasculitis, multiple sclerosis, Bechet) (n=19), mild cognitive impairment (MCI) (n=2), and others {dystonia, polyneuropathy, ataxia telangiectasia (ATM)} (n=4);

Psychiatric diseases: (depression, psuedodementia) (n=28). Demographic data included gender, age at onset and the time point of lumbar-puncture (LP). When available, scores of the Mini Mental State Examination (MMSE) were reported. In a small part of the study population, the number of years of education or the type of occupation (divided into 3 degrees of challenging level) was available. The CSF samples of these patients were sent to our laboratory as a routine work-up for dementia. The diagnosis, made according to medical data and follow-up using established internationally agreed criteria, was based on medical records collected from the different medical centers. The results of the CSF T-tau analysis were not included in the diagnostic criteria. In the AD group, although the diagnosis was based on international clinical criteria (clinically probable AD) - it may be assumed that the mixed pathology of AD with VD is quite common. For comparison, we also included the group affected with Creutzfeldt-Jacob disease (CJD) (n=76) patients, previously reported by us [24]. Clinical data from the patients is registered in The Israeli National Registry Database of CJD overseen by one of us (E.K) according to Israeli’s laws. Maintenance of participants’ anonymity was strictly kept during database analyses. We summarized here the results of CSF samples tested by us in 1999-2007. Clinical follow-up was performed during at least three years allowing confirmation of diagnosis.

CSF analysis

CSF samples were tested for T-Tau protein at our Laboratory of Neurogenetics, a national referral center of the Department of Neurology, Hadassah Hebrew University Medical Center. This laboratory is a national referral center for genetic and CSF analysis for neurodegenerative diseases in Israel. CSF samples were sent to us (usually within 24 to 48 hours following LP stored in plastic tubes). Tau concentrations were measured by ELISA (Innotest hTau-Ag, Innogenetics, Ghent, Belgium).

Statistical analysis

To compare continuous variables among all of the patient groups, the non-parametric Kruskal-Wallis Test was used. To compare continuous variables between two groups, the two sample t-test and the non-parametric Mann-Whitney test were applied. For multiple pair-wise analyses, the Bonferroni correction of the significance level was used, as follows: p<0.0035 for both age and tau levels; p=0.01 for MMSE scores. Comparisons were performed only in groups with n≥8 patients. The comparison of qualitative variables between two groups was carried out using the Chi-Square test. To assess linear associations between two continuous variables, the Pearson correlation coefficient was calculated. Analysis of covariance (ANCOVA) was used for simultaneously assessing the effect of several independent variables (both continuous and qualitative) on a dependent continuous variable. Sensitivity, specificity and positive and negative predicting values were calculated. Receiver Operating Characteristic curve (ROC) analysis was performed in order to find the optimal cut-off point of CSF tau for differentiation between AD patients and other patient groups {with a relatively high number of cases (n≥28): VD, FTD, psychiatric diseases, CJD}. All of the applied tests were two-tailed, and a p-value of 5% or less was considered statistically significant.



The demographic data and the MMSE scores of the study population are summarized in the Table. Comparisons were performed for groups containing at least eight subjects. The mean age of disease onset in the AD patients was 60.02±8.68 (M ± SD), and the age at lumbar puncture (LP) was 61.95±8.32. Comparing these ages with those of other diseases (two sample t-test) revealed a significantly different age (higher) only in the VD group (67.06±10.37, and 67.84±10.41, respectively, p<0.0001 for each). Some differences in ages relative to the AD group were also noticed in other groups: higher in mixed dementia (67.17±8.85, 68.74±8.81, respectively), while lower in the encephalitis group (43.52±23.11, 45.99±27.52), CVA (onset: 67.09±13.31), autoimmune diseases (48.67±15.45, and 50.75±15.18) and psychiatric patients (54.12±12.97, 55.41±12.73); however, Bonferroni correction excluded significant differences. No significant difference in the male to female ratios (M/F) was detected between the AD patients (60/64) and the other groups. Some differences in mean MMSE scores were noticed (two sample t-test) (18.17±6.49 in the AD group, while 21.53±5.89 and 23.38±5.78 was noted in the FTD and the psychiatric group, respectively), yet Bonferroni correction excluded a significant difference.