Advances in the detection of Alzheimer's disease—use of cerebrospinal fluid biomarkers

Abstract

The diagnosis of Alzheimer's disease (AD) is still made by excluding other disorders with a similar clinical picture. In addition, an analysis of symptoms and signs, blood analyses and brain imaging are the major ingredients of the clinical diagnostic work-up. However, the sensitivity of a clinical diagnosis using these instruments is unsatisfactory and disease markers with high sensitivity and specificity for AD would be a welcome supplement. Ideally, such markers should reflect the pathophysiological mechanisms of AD, that is, according to the currently predominant hypothesis mismetabolism of β-amyloid and neurofibrillary degeneration. Among several, we have focused on three candidates that have been suggested to fulfil the requirements for biomarkers of AD: β-amyloid42 (Aβ42), total tau (T-tau) and tau phosphorylated at various epitopes (P-tau). The cerebrospinal fluid (CSF) levels of these proteins reflect the metabolism of these proteins in the central nervous system. Only published articles using established ELISA methods for the quantification of these markers in CSF and preferably also presenting sensitivity and specificity figures have been included in this review. The number of patients included in the different studies varies; some having included only a few patients. Furthermore, diagnostic criteria vary and clinicopathological studies are scarce. However, there are some large studies, including even minor studies, and most have found reduced CSF levels of Aβ42 and increased CSF levels of T-tau in AD. The sensitivity and specificity of these measures are high for separation of AD patients from controls, but their specificity against other dementias is moderate. It increases if P-tau is added. An increasing number of studies suggest that supplementary use of these CSF markers, preferably in combination, adds to the accuracy of an AD diagnosis.

Introduction

Alzheimer's disease (AD) is a major health threat and one of the most costly diseases in modern society. Estimates indicate that approximately 4 million people in the United States suffer from AD. The disorder, which affects men and women equally, is characterized by progressive deterioration of cognitive functions, such as memory, language and visuospatial orientation. Associated symptoms are mood and behavioural changes. The prognosis is poor with no cure available. AD imposes a heavy burden on the caregivers in the families and in the health care system. Society's total cost for care of patients with dementia is substantial. Estimates have suggested that the cost for a country such as Sweden is about 0.4 billion Euro per million inhabitants [1]. A rapid increase in the elderly population is expected, especially in the developing countries, and this will lead to a further increase in the cost, if no cure is found.

The cause of AD is unknown in most cases, but mutations in a few underlying genes have been identified in familial AD [2]. Underlying neuropathological changes in AD are the accumulation of senile plaques (SPs) and neurofibrillary tangles (NFTs). SPs are made up mainly of β-amyloid, especially the 42-amino-acid isoform, β-amyloid 42 (Aβ42) [3]. The major constituent of NFTs is a cytoskeleton-associated protein called tau, which is hyperphosphorylated in NFTs [4]. The golden standard of diagnosis is the identification of typical neuropathological changes in the brain of a patient who has suffered from clinical AD. In clinical routine, the diagnosis of AD, as outlined in the NINDS-ADRDA criteria [5], is based on clinical and neuropsychological examinations, identification of typical symptoms of AD and exclusion of other known causes of dementia. Studies have shown that the accuracy of the clinical diagnosis is between 65% and 90%. Higher accuracy is achieved at academic centres with special interest in AD. This is partly due to the fact that these centres include patients in the later stages of the disease who have been followed for several years before the confirming autopsy [6], [7], [8]. The accuracy of the clinical diagnosis at the primary care level and in general hospitals is probably even lower, especially in the early stages of the disease when the symptoms are indistinct. In view of this, the need for specific AD markers is great. According to a proposal of a consensus group on molecular and biochemical markers of AD [9], an ideal marker of AD should be able to detect a fundamental feature of neuropathology and should be validated against neuropathologically confirmed cases. Furthermore, its sensitivity for detection of AD as well as its specificity for discrimination of AD from other dementia disorders should exceed 80%. A marker for AD should also be reliable, reproducible, noninvasive, simple to perform in clinical routine and inexpensive.

As the cerebrospinal fluid (CSF) is in direct contact with the extracellular space of the brain [10], biochemical changes in the brain, for instance, those caused by accumulation of SPs and NFTs, will lead to a change in the biochemistry of the CSF [11]. CSF is therefore an appropriate source of biochemical markers for AD. There are three candidates that have been suggested to fulfil the requirements stated by the consensus group [9]: total tau protein (T-tau), Aβ42 and tau phosphorylated at AD-specific epitopes (P-tau).

Tau is a microtubule-associated protein which is located mainly in neuronal axons. By binding to microtubules, it promotes the stability and function of these. In the normal human brain, six different isoforms of tau are found, all of which have numerous phosphorylation sites [12]. As tau is a major constituent of NFTs, CSF T-tau has been suggested as a marker for AD. Using monoclonal antibodies that detect all isoforms of tau independent of degree of phosphorylation, enzyme-linked immunosorbent assays (ELISAs) have been developed that measure the T-tau levels in CSF [13], [14], [15] (Fig. 1).

Using these ELISAs, more than 50 studies have consistently demonstrated a moderate to marked increase in CSF T-tau as well as high sensitivity and specificity of CSF-tau in AD patients when compared with controls (Table 1). So far, CSF from about 2400 AD patients and 1250 controls has been investigated in this way (Table 1). The mean degree of increase is about 300% in AD compared with controls. The high sensitivity and specificity make CSF T-tau a good candidate for the designation biochemical marker for AD, or AD biomarker. However, high levels of T-tau in the CSF have also been found in a proportion of cases with other dementia disorders, such as frontotemporal dementia [16], [17] and Lewy body dementia [18], but in several other disorders, for example, alcohol dementia, Parkinson's disease and depression, the CSF levels of T-tau seem to be normal and only occasionally increased [14], [17], [19], [20].

What does an increase in CSF T-tau reflect? Few studies have directly investigated this, but it has been suggested that the CSF T-tau levels reflect the degree of neuronal (especially axonal) degeneration and damage [14]. Some evidence for this has been found, for instance, a transient increase in CSF T-tau after acute stroke, with a positive correlation between CSF T-tau and infarct size as measured by computerized tomography [21], a very marked increase in CSF T-tau in Creutzfeldt–Jakob's disease [22], and a correlation between premortem CSF T-tau levels and the post-mortem density of neurofibrillary tangles in the brain [23]. Indirect evidence is that, in AD and controls, there is a positive correlation between the CSF levels of T-tau, GAP-43 and amyloid precursor protein (APP), all proteins located in the axon of neurons [24].

The central protein in SPs is Aβ42. It is produced and secreted from human cells as a result of normal cellular processing of the larger transmembrane protein APP [25] (Fig. 2). In this processing, APP is cleaved in several steps and Aβ is produced. In, AD, APP is first cleaved by an enzyme called β-secretase, which results in the release of a large N-terminal fragment called β-secretase-cleaved soluble APP. In a second step, APP is cleaved by the γ-secretase complex, which results in the release of free Aβ (Fig. 2). In this processing, various isoforms of Aβ, for example, Aβ42, are produced; all of which are secreted into the CSF.

Using four different ELISA methods that are specific to Aβ42 [26], [27], [28], [29], more than 30 studies have consistently demonstrated a moderate to marked decrease in CSF Aβ42 in AD. The principle for the ELISA that is most commonly used to measure Aβ42 in CSF, INNOTEST™ β-AMYLOID_(1–42) [29], is shown in Fig. 2. There are 13 studies, including a total of about 600 AD cases and 450 controls, in which sensitivity and specificity figures have been given or can be calculated from graphs (Table 2). These studies show that, for CSF Aβ42, the mean sensitivity for discrimination between AD and normal aging is approximately 86%, while the specificity is approximately 91% and the mean level of decrease in AD patients compared with controls is about 50% (Table 2). On the other hand, the specificity for discrimination of AD from other disorders is moderate. Low levels of Aβ42 in CSF have, for example, been found in Lewy body dementia [18], [30], a disorder also characterized by the presence of SPs. Low levels have also been found in a small percentage of patients with frontotemporal dementia and vascular dementia [31], [32] and also in Creutzfeldt–Jakob's disease [33], [34] and amyotrophic lateral sclerosis [35]. These studies question the putative relation between low CSFAβ42 levels and the accumulation of SPs. There are several possible causes of low CSF-Aβ42 levels, for example, axonal degeneration [35], [36] and entrapment in narrow interstitial and subarachnoid drainage pathways [37].

Tau is normally in a phosphorylated state. Over 70 phosphorylation sites are found on the human tau molecule (Fig. 3) and, in AD, tau is usually in a hyperphosphorylated state. In AD, this hyperphosphorylation involving certain epitopes on the tau molecule has the consequence that tau loses its ability to promote microtubule assembly and stability, which in turn leads to cytoskeleton instability and diminished transport ability [38], [39]. A consequence of this is aggregation of tau with subsequent formation of NFTs [12]. Several ELISAs have been developed that use monoclonal antibodies directed toward sites that are phosphorylated in AD. The principle for one of these ELISAs, INNOTEST™ PHOSPHO-TAU_(181P), which measures tau phosphorylated at threonine 181 (P-Tau₁₈₁), is given in Fig. 3 [40]. Other ELISAs identify tau phosphorylated at the epitopes threonine 181 and 231 (P-tau_181+231) [14], threonine 231 and serine 235 (P-tau_231+235) [41], serine 199 (P-tau₁₉₉) [41], threonine 231 (P-tau₂₃₁) [42] and serine 396 and 404 (P-tau_396+404) [43].

All these assays have shown increased CSF levels of P-tau in AD patients compared with controls (Table 3). The sensitivity of CSF P-tau for discrimination between AD and normal aging is about the same or slightly lower as that of CSF T-tau, that is, about 75%. Interestingly, the specificity of CSF P-tau for discrimination of AD from other dementias seems to be higher than those of CSF T-tau and CSF Aβ42. Normal CSF levels of P-tau have been found in vascular dementia, frontotemporal dementia [44] and Lewy body dementia [45], which suggests that the above ELISAs may help to discriminate between AD and these dementias. In addition, while there is a marked increase in CSF T-tau after acute stroke, the CSF P-tau does not change [46]. This suggests that the origin of increased CSF P-tau levels is more closely related to AD pathology, for instance, the formation of NFTs.