Inflammatory Proteins in Plasma and the Risk of Dementia: Title and subTitle BreakThe Rotterdam Study FREE

Neuritic plaques and neurofibrillary tangles are the classic neuropathological features in the brains of patients with Alzheimer disease (AD). Furthermore, inflammation has been demonstrated in AD-affected brains as is indicated by the presence of activated microglia and inflammatory proteins.¹ In addition to inflammation within the brain, the peripheral immune system of patients with dementia may be activated as well. This is suggested by cross-sectional studies that reported elevated blood levels of the inflammatory proteins α1-antichymotrypsin (ACT),^{2 - 4} C-reactive protein (CRP),⁵ (IL) 6⁶ in patients with AD compared with control subjects. Prospectively, increased CRP levels have been associated with an increased risk of dementia 25 years later.⁷ Whether the levels of ACT and IL-6 are increased before the onset of dementia is yet unknown, although the recent finding of an association between the level of IL-6 and cognitive decline suggests elevated IL-6 levels before dementia.^{8 - 9} To our knowledge, no studies examined the association between the levels of ACT and IL-6 and dementia in a population-based setting, nor has the association been studied between the levels of other inflammatory proteins, such as cell adhesion molecules (CAMs), and dementia. We investigated the relation between plasma ACT, CRP, IL-6, soluble intercellular adhesion molecule-1 (sICAM-1), soluble vascular cell adhesion molecule-1 (sVCAM-1), and the risk of dementia in a large population-based prospective study.

The Rotterdam Study is a population-based prospective cohort study among persons aged 55 years and older, who were living in Ommoord, a suburb of the city of Rotterdam, the Netherlands.¹⁰ The study was approved by the medical ethics committee of the Erasmus Medical Center, Rotterdam. The objective was to investigate determinants of chronic and disabling diseases.¹⁰

The overall response rate was 78% corresponding to 7983 participants, from whom written informed consent was obtained. At the baseline examination (1990-1993) blood samples were drawn from 7050 subjects (88% of the total cohort). Of these, cognitive status was assessed in 7047 subjects. Dementia was diagnosed in 334 participants resulting in a source population of 6713 subjects who both were dementia free and had venipuncture at baseline. Follow-up examinations occurred in the time spans of 1993-1994 and 1997-1999.

The study population comprises a subcohort of 727 subjects who were randomly drawn in July 1998 from the source population. At that time, we additionally selected all 146 cases of incident dementia that had occurred outside the subcohort. We followed up the subcohort until December 31, 1999. By then, 42 subjects of the subcohort had developed dementia. Thus, cases of the present study were composed of 188 cases of incident dementia, in which 140 subjects were diagnosed as having AD, 23 subjects as having vascular dementia, and 25 subjects as having other dementias.

During baseline and follow-up examinations the diagnosis of dementia followed a similar 3-step protocol.¹¹ Two brief tests of cognition (Mini-Mental State Examination¹² and Geriatric Mental State Schedule organic level¹³ ) were used to screen all subjects. Participants with positive screening scores of less than 26 on the Mini-Mental State Examination or an organic level of more than 0 on the Geriatric Mental State Schedule underwent further cognitive testing using the Cambridge examination (CAMDEX) for mental disorders of elderly persons.¹⁴ Subjects who were thought to have dementia were examined by a neurologist (J.C.vS.) and a neuropsychologist; if possible, the brains of these subjects also underwent magnetic resonance imaging. In addition, the total cohort was continuously monitored for incident dementia through computerized linkage between the study database and digitalized medical records from general practitioners and the Regional Institute for Outpatient Mental Health Care, Rotterdam.¹¹ Dementia diagnoses were made in accord with internationally accepted criteria for dementia (DSM-III-R),¹⁵ AD (National Institute of Neurological and Communicative Disorders and Stroke–Alzheimer's Disease and Related Disorders Association [NINCDS-ADRDA]),¹⁶ and vascular dementia (National Institute of Neurological Disorders and Stroke–Association Internationale pour la Recherche et l'Enseignement en Neurosciences [NINDS-AIREN]).¹⁷

The levels of inflammatory proteins from the study population were assessed in baseline plasma samples that had been stored at −80°C. The ACT and CRP levels were measured by kinetic nephelometry (Nephelometer BN200; Dade-Behring, Marburg, Germany). The CRP measurement is highly sensitive and the World Health Organization CRP Reference Standard was used. The levels of IL-6, sICAM-1, and sVCAM-1 were determined by means of commercially available enzyme-linked immunosorbent assays (Quantikine HS IL-6 kit; R & D Systems Europe, Oxon, England; sICAM-1 and sVCAM-1 kits; R & D Systems Europe). Interassay coefficients of variations for ACT, CRP, IL-6, sICAM-1, and sVCAM-1 were 2.8%, 4.4%, 8.7%, 6.9%, and 5.0%, respectively.

At baseline we obtained information on the subject's educational level, tobacco smoking, and use of anti-inflammatory medication (nonsteroidal anti-inflammatory drugs [NSAIDs] or corticosteroids) or statins. In addition, we determined body mass index (calculated as weight in kilograms divided by the square of height in meters), systolic blood pressure, and the presence of diabetes mellitus. An ultrasonogram of both carotid arteries was performed.

As indicators of atherosclerosis, we used the following 3 measures: peripheral arterial disease defined as an ankle-brachial ratio of the systolic blood pressure below or above 0.9, intima media thickness of the carotid arteries,¹⁸ and atherosclerotic carotid plaques determined at 6 different locations¹⁸ according to which 4 categories were created: 0, 1 to 2, 3 to 4, and 5 to 6.

The association between plasma levels of ACT, CRP, IL-6, sICAM-1, and sVCAM-1, respectively, and incident dementia and its subtypes was evaluated in a case-cohort design by use of standard Cox proportional hazards regression models with modification of standard errors based on robust variance estimates.¹⁹

The levels of inflammatory proteins were first entered into the model as quintiles to explore any deviations from linearity in the relation with dementia. In case of a nonlinear association or a threshold-effect, all analyses were performed using quintiles. Otherwise, the levels of inflammatory proteins were entered into the model through a linear term, in which the regression coefficient was expressed per standard deviation increase. All analyses were adjusted for age, gender, and educational level, and additionally for tobacco smoking, body mass index, diabetes mellitus, use of anti-inflammatory medication, and atherosclerosis. The analyses were repeated after exclusion of subjects taking anti-inflammatory medication (n = 93, 10.7%), subjects with possible acute inflammation as is indicated by CRP levels higher than 15 mg/L (n = 53, 6.1%), and users of statins (n = 18, 2.1%), respectively. Analyses were performed using SAS statistical software (version 6.12; SAS Institute Inc, Cary, NC).

Table 1 gives the baseline characteristics of the subcohort. The mean (SD) age was 71.7 (9.0) years, a slight majority were women, and one third had only a primary education.

In Figure 1 the median level of inflammatory proteins within quintiles was plotted against rate ratios of dementia on a logarithm scale. The rate ratio of the highest quintile compared with the lowest quintile was 2.92 (95% confidence interval [CI], 1.34-6.36) for ACT, 1.37 (95% CI, 0.67-2.77) for CRP, 1.42 (95% CI, 0.62-3.23) for IL-6, 1.20 (95% CI, 0.47-3.08) for sICAM-1, and 0.55 (95% CI, 0.27-1.13) for sVCAM-1. From the Figure 1, we saw no reason to assume that the associations could not be described by a linear term given the wide and overlapping 95% CIs of the point estimates within quintiles.

Table 2 lists the rate ratios of dementia and its subtypes per standard deviation increase in the level of inflammatory protein, adjusted for age, gender, and educational level. Higher levels of ACT, IL-6, and, to a lesser extent, CRP were associated with an increased risk of dementia. Rate ratios were 1.49 (95% CI, 1.23-1.81) for ACT, 1.12 (95% CI, 0.99-1.25) for CRP, and 1.28 (95% CI, 1.06-1.55) for IL-6. The levels of ACT, CRP, and IL-6 were also associated with an increased risk of AD, although the levels of ACT and CRP were most strongly associated with vascular dementia. The levels of sICAM-1 and sVCAM-1 were not associated with dementia and its subtypes.

Data in Table 3 demonstrate that the relation between the levels of inflammatory proteins and dementia did not change substantially after additional adjustments for tobacco smoking, body mass index, diabetes mellitus, use of anti-inflammatory medication, or atherosclerosis.When the analyses were performed after exclusion of the data for users of anti-inflammatory medication, subjects with CRP levels higher than 15 mg/L, or users of statins, the results also remained similar.

We observed that elevated plasma levels of ACT and IL-6 were associated with an increased risk of dementia, AD, and vascular dementia. This association was less clear for CRP and was nonexistent for sICAM-1 and sVCAM-1.

Strengths of the present study are its population-based design, size, complete dementia follow-up, and the adjustments for an extensive number of confounders. A limitation of the study is the lack of repeated measurements of the levels of inflammatory proteins, which may have resulted in dilution of the associations owing to within-person variability and measurement error. In addition, we lack data on white matter lesions and small infarcts in the brain. As a consequence, we do not know whether white matter lesions and small infarcts may explain the observed association between the level of inflammatory proteins and dementia.

Our findings agree with cross-sectional studies showing that blood levels of ACT and IL-6 are higher in patients with dementia compared with controls.^{2 - 4 ,6} In addition, our observations on CRP are in accord with the recent finding of elevated CRP levels 25 years before the clinical onset of dementia, AD, and vascular dementia.⁷ To our knowledge, blood levels of CAMs in relation to dementia have not been studied before, but neuropathological findings showed no expression of these adhesion molecules on the endothelial cells of capillaries in AD-affected brains.²⁰

Elevated plasma levels of ACT and CRP before the diagnosis of dementia likely originate from an increased production in the peripheral immune system, because the levels of ACT and CRP in brains are generally more than 100 times lower than in plasma.³ In contrast, increased levels of IL-6 may be derived from increased production either in the brain or in the periphery, because IL-6 levels are similar in brain and plasma.⁶

Increased levels of inflammatory proteins in plasma before the clinical onset of dementia may suggest that peripheral inflammation is involved in the pathogenetic process leading to dementia. This is supported by several observations. First, anti-inflammatory medication use, such as NSAID use, is associated with a decreased risk of dementia.^{21 - 22} Although NSAIDs can have affects within the brain, the effect of NSAIDs on dementia risk may also act by suppressing the peripheral immune system. Second, ACT is known to reinforce the formation of β-amyloid deposits,^{23 - 24} which is thought to be crucial in the pathogenesis of AD. Because peripheral proteins are able to cross the blood-brain barrier²⁵ and because the permeability of the blood-brain barrier may increase with rising plasma levels of inflammatory proteins such as IL-6,²⁶ peripherally produced inflammatory proteins might also amplify formation of cerebral β-amyloid deposits. Third, atherosclerosis, which is an inflammatory process,²⁷ may be associated with increased risk of dementia.¹⁰ The idea of atherosclerosis being a link between peripheral inflammation and dementia is supported by our observation of relatively strong associations of ACT and CRP with vascular dementia. However, adjustment for atherosclerosis did not change our results and the risk associated with IL-6 was similar for AD and vascular dementia. Also, because atherosclerosis may be associated with increased plasma levels of sICAM-1,¹⁸ one would also have expected an association between sICAM-1, sVCAM-1, and the risk of dementia if atherosclerosis was to contribute to the pathogenesis of dementia. On the other hand, elevated levels of inflammatory proteins in plasma may be a consequence of the pathophysiological process of dementia. Peripheral inflammation may result from cerebral β-amyloid deposition that induces local production of inflammatory proteins, such as IL-1 and IL-6.¹ These cerebrally produced cytokines may increase peripheral levels of inflammatory proteins because they pass either through the blood-brain barrier²⁵ or through stimulation of peripheral production of inflammatory proteins as shown in animal models.²⁸ If the activation of the peripheral immune system is both a cause and a consequence of the pathogenetic process of dementia, a self-enhancing cascade will occur. This cascade includes β-amyloid deposit formation that leads to local inflammation within the brain resulting in the activation of the peripheral immune system that leads to increased β-amyloid deposit formation.

High plasma levels of ACT, IL-6, and, to a lesser extent, CRP were associated with an increased risk of dementia, AD, and vascular dementia. Whether peripheral inflammation contributes to the pathogenesis of dementia remains to be elucidated.

Corresponding author: Monique M. B. Breteler, MD, PhD, Department of Epidemiology and Biostatistics, Erasmus Medical Center, PO Box 1738, 3000 DR Rotterdam, the Netherlands (e-mail: m.breteler@erasmusmc.nl).

Accepted for publication December 29, 2003.

Author contributions: Study concept and design (Drs Engelhart, Geerlings, Kiliaan, Ruitenberg, Hofman, and Breteler); acquisition of data (Drs Engelhart, Kiliaan, Ruitenberg, van Swieten, and Breteler and Mr Meijer); analysis and interpretation of data (Drs Engelhart, Geerlings, Ruitenberg, van Swieten, Stijnen, Witteman, and Breteler); drafting of the manuscript (Drs Engelhart, Geerlings, Kiliaan, and Breteler and Mr Meijer); critical revision of the manuscript for important intellectual content (Drs Engelhart, Geerlings, Ruitenberg, van Swieten, Stijnen, Hofman, Witteman, and Breteler); statistical expertise (Drs Stijnen, Witteman, and Breteler); obtained funding (Drs Kiliaan, Hofman, and Breteler); administrative, technical, and material support (Dr Kiliaan and Mr Meijer); study supervision (Drs Geerlings, Ruitenberg, Hofman, and Breteler).

This study was supported by the Netherlands Organization for Scientific Research, The Hague, by an unrestricted grant from Numico Research BV, Wageningen, the Netherlands, and by grant IIRG-99-1534 from the Alzheimer's Association, Maastricht, the Netherlands (Dr Breteler).