Inflammation,Antiinflammatory Agents,and Alzheimer’s Disease: The Last 22 Years

INTRODUCTION

Two basic discoveries led to the development of neuroinflammation as a major field of brain research. The first was the immunohistochemical demonstration of reactive microglia in Alzheimer’s disease (AD) brains [1, 2], and the second was that persons suffering from rheumatoid arthritis had a greatly reduced risk of developing AD [3]. The first was interpreted as indicating the existence of a chronic inflammation in AD brain, while the second was considered to be a beneficial consequence after use of non-steroidal anti-inflammatory drugs (NSAIDs) prior to clinical disease manifestation [3]. In the ensuing years, there have been over 4,000 reports expanding the evidence of chronic inflammation in AD brain. There are now a large number of reviews on the subject (e.g., [4]). Reactive microglia have been shown to produce free radicals and other neurotoxic substances that kill neurons in culture [5, 6]. Some activated T cells are also found in the brain parenchyma in AD [7–9]. Such cells release inflammatory mediators, including the powerful proinflammatory stimulants interleukin (IL)-1, IL6, tumor necrosis factor alpha (TNF-α), and γ-interferon. Targeting TNF has been considered as a therapeutic strategy for AD [10].

A key discovery linking inflammation to AD was the finding that aggregated amyloid-β (Aβ) alone was a powerful activator of complement [11]. It had previously been shown that the complement system was activated in AD [12], but antibodies, which were then considered to be the main activators of complement, could not be identified in association with AD lesions.

The complement system has now been shown to be fully activated in AD [13]. Once activated, it produces anaphylatoxins, which promote further inflammation. Its opsonizing components mark material for phagocytosis. If fully activated, the membrane attack complex (MAC) is directly lytic to cells (Fig. 2). The MAC inserts itself into viable cell membranes, causing them to leak with subsequent death. It is intended to destroy foreign cells and viruses, but host cells are at significant risk in a phenomenon known as bystander lysis.

Figure 2 illustrates the difference between opsonization by the early complement components and the lytic effects of the MAC. Figure 2a shows double immunostaining for C4d, a fragment of the opsonizing pathway of complement, and complement receptors. C4d attaches covalently to the amyloid deposit, while activated microglia, which express high levels of complement receptors, are attacking the deposit by attraction to their ligand. Figure 2b shows a different phenomenon that is invisible to the stains used in Fig. 2a. This is the attack on neurites within the plaque by the terminal complement components C5b-9, which require a viable membrane for MAC assembly. The MAC has a very short half-life so finding immunohistochemical evidence for its existence in postmortem AD brains indicates the vigor of the attack. A more revealing overall index may be the levels of RNA expression of the complement proteins. There is a marked upregulation in affected regions in AD [14] (Fig. 3).

Identification of the MAC attacking dystrophic neurites in AD [13, 15] provides strong evidence of self attack in AD. It also provides the only in vivo evidence linking Aβ deposits with neurotoxicity. A host of other inflammatory markers have now also been shown to be upregulated in affected brain areas in AD. They include many of the inflammatory cytokines and such inflammatory stimulants as ICAM-1. Alleles that favor production of IL-1α, IL-1β, and IL-8, as well as TNF and other inflammatory cytokines, have been frequently reported to increase the risk of AD [16].

In the past 25 years, there have been more than 15 epidemiological studies showing that individuals are relatively spared from AD if they have been taking NSAIDs, or have suffered from conditions where such drugs are routinely used [4, 17–21]. Four large epidemiological studies have analyzed the effects of NSAID consumption on AD. The Baltimore longitudinal study [18] showed a sparing of approximately 60% among NSAID users of greater than 2 years duration; the Cache County study showed a sparing of approximately 55% [21]; the Rotterdam study, where NSAID consumption was verified through prescription records, showed an 80% sparing [19]; and the MIRAGE study showed a sparing of 36% [20]. Some NSAIDs at very high doses directly bind to Aβ, but Szekely et al. showed they were no more effective than other NSAIDs in reducing the risk of AD [22].

BIOMARKER STUDIES

Biomarker studies have opened up a new era for AD research. So far there are three reliable types: cerebrospinal fluid (CSF) to determine Aβ and tau secretion levels, positron emission tomography (PET) to determine Aβ deposit levels (Pittsburgh Compound B, PIB), PET to determine metabolic rate (FDG), and MRI to determine brain volume. Overall, they show that AD onset occurs a decade or more before clinical symptoms appear.

Bateman et al. [23], in a landmark study of 128 participants, found that concentrations of Aβ₄₂ in the CSF declined 25 years before the expected clinical onset. Aβ deposits in the brain, as revealed by PIB, were detected 15 years before onset. They were concomitant with increased tau in the CSF and an increase in brain atrophy. Impaired episodic memory was observed ten years before the expected clinical diagnosis, and declines in the Mini-Mental State Examination and the Clinical Dementia Rating scale were detected 5 years before the expected clinical diagnosis.

Comparable findings were reported by Villemagne et al. [24], who estimated that it took 19.2 years of linear Aβ accumulation, 4.2 years of hippocampal atrophy, and 3.3 years of memory impairment to reach AD clinical diagnostic levels. Seppala et al. [25] correlated CSF findings with cortical biopsy analysis and found that patients with Aβ cortical plaques in biopsy samples had lower Aβ₄₂ CSF levels than those without plaques. Prestia et al. [26], following patients with minimal cognitive impairment (MCI), found that conversions to dementia increased as patients progressed from the appearance of Aβ₄₂ in the CSF, to abnormal Aβ₄₂ and FDG by PET, to hippocampal atrophy with Aβ₄₂ and FDG by PET.

Okonko et al. [27] found that abnormal Aβ₄₂ in the CSF, but not tau alterations, were associated with increased risk of AD, and Buchhave et al. [28] reported similar results in a study of patients with MCI followed for a median of 9.2 years. They concluded that 90% of patients with MCI and pathologic CSF biomarkers develop AD within 9 to 10 years, and that Aβ₄₂ is being deposited 5–10 years before the appearance of dementia. Shaw et al. [29] measured CSF biomarkers in mild AD and MCI patients compared with controls, as well as autopsy confirmed cases compared with controls. They concluded that Aβ₄₂ plus total tau predicted conversion of MCI to AD and that Aβ₄₂ was the most sensitive marker in the autopsy cases. Similarly, Visser et al. [30], in the DESCRIPA study involving a prospective cohort, found that patients with an AD profile in their CSF were prone to advance from MCI to AD type dementia.

It is important to recognize that there is a fundamental difference between CSF and brain biomarkers. CSF turns over every 4–6 hours so CSF biomarkers provide a differential measure, revealing only the rate of production during that brief period of time. Brain biomarkers have a cumulative effect, so they are integral markers representing years of activity. Zetterberg et al. [31] found that CSF biomarker production remained constant over a two-year period as cognitive activity declined. Mattson et al. [4] found CSF biomarker production to be constant over a four-year period in MCI patients. Both these groups reached the conclusion that a CSF profile shifting toward normal would be useful in tracking disease-modifying drugs.

As for Aβ deposits in the brain, Vlassenko et al. [32] found that scans with PIB about 2 years apart in cognitively normal adults showed that those with elevated binding also showed enhancement of such binding, indicative of increased brain Aβ deposits. They concluded that a major growth in the Aβ burden occurs during a preclinical stage of AD. Bruck et al. [33] compared the prognostic value of PIB-PET, FDG-PET, and hippocampal volume MRI, for their prognostic value in predicting conversion of MCI to AD. Of the 29 patients, 17 converted to AD after 2 years. They concluded that the PET methods were superior to hippocampal volume methods in predicting the conversion. Hatashita and Yamasaki [34] followed 68 MCI patients by PIB-PET and FDG-PET. Over 19 months, 44% of the patients converted to AD. They found PIB-PET to be the most definitive marker of MCI. Jack et al. [35] have proposed a model in which Aβ biomarkers become abnormal first, with neurodegenerative biomarkers becoming abnormal later, correlating with clinical symptom severity. To apply these findings on a widespread basis to inhibit AD development, a simple, cheap, non-invasive test of disease onset is required.

CONCLUSIONS

In summary, biomarker data indicate that AD onset can be detected at least ten, and possibly 20 years prior to clinical diagnosis. They suggest that an extended window of opportunity exists for appropriate AD therapy to ameliorate, or even to prevent disease development. Such therapy would involve administration of NSAIDs and other anti-inflammatory agents. Widespread application will require development of a simple and reliable diagnostic method.