- Open Access
The association of high sensitivity C-reactive protein and incident Alzheimer disease in patients 60 years and older: The HUNT study, Norway
Immunity & Ageingvolume 15, Article number: 4 (2018)
With ageing, long-standing inflammation can be destructive, contributing to development of several disorders, among these Alzheimer’s disease (AD). C-reactive protein (CRP) is a relatively stable peripheral inflammatory marker, but in previous studies the association between highly sensitive CRP (hsCRP) and AD have shown inconsistent results. This study examines the association between AD and hsCRP in blood samples taken up to 15 years prior to the diagnoses of 52 persons with AD amongst a total of 2150 persons ≥60 years of age.
Data from Norway’s Nord-Trøndelag Health Study (HUNT 2) and the Health and Memory Study (HMS) were linked. The participants had an average age of 73 years, and diagnosed with AD up to 15 years [mean 8.0 (±3.9)] following hsCRP measurement. Logistic regression models showed an adverse association between hsCRP and AD in participants aged 60-70.5 (odds ratio: 2.37, 95% CI: 1.01-5.58). Conversely, in participants aged 70.6-94, there was an inverse association between hsCRP and AD (odds ratio: 0.39, 95% CI: 0.19-0.84). When applying multivariate models the findings were significant in individuals diagnosed 0.4-7 years after the hsCRP was measured; and attenuated when AD was diagnosed more than seven years following hsCRP measurement.
Our study is in line with previous studies indicating a shift in the association between hsCRP and AD by age: in adults (60-70.5 years) there is an adverse association, while in seniors (>70.6 years) there is an inverse association. If our findings can be replicated, a focus on why a more active peripheral immune response may have a protective role in individuals ≥70 years should be further examined.
Pre-clinical and clinical studies have shown that the immune system contributes and drives Alzheimer disease (AD) pathogenesis . Inflammatory proteins found outside of the brain have also been shown to be elevated in patients with AD . With ageing, long-standing inflammation can be destructive, contributing to development of several disorders [3, 4].
A minor elevation in inflammatory markers in blood is termed low-grade inflammation, where the body is constantly under very mild chronic elevation, but not to the extent of acute inflammation . Low grade inflammation is recognized as an important contributor to the pathophysiology of hypertension, to the initiation and progression of atherosclerosis and the development of cardiovascular disease (CVD) . C-reactive protein (CRP) is a relatively stable peripheral inflammatory marker that has been used as a marker of low-grade inflammation, and the highly sensitive assay (hsCRP) has been shown to be moderately elevated in acute myocardial infarction, coronary artery disease, metabolic syndrome, neurodegenerative diseases, and hypertension [7,8,9,10]. Since several CVD have been shown to share risk factors for developing dementia, a number of studies have examined whether there is an association between low-grade peripheral inflammation and AD [11, 12]. However, previous epidemiological studies examining hsCRP and AD revealed conflicting findings. Studies examining hsCRP during midlife showed adverse associations, where moderate elevations of hsCRP were increased in persons who developed AD later in life . In contrast, studies examining older participants published that higher plasma levels of hsCRP was associated with a lower risk for AD and all-cause dementia, and authors questioned whether this could be attributed to a genetic phenotype for successful aging [14, 15]. Some studies examining gene expression have shown down-regulation of immune response genes in brain regions of cognitively impaired oldest-old persons and up-regulation in cognitively intact individuals of same age [16, 17]. Locascio et al. found that low levels of hsCRP were associated with more rapid progression of illness, whereas Nilsson et al. found that although CRP was overall lower in persons with AD, elevated CRP was associated with shorter survival time [18, 19]. Other studies show the opposite, that high hsCRP levels were associated with cognitive decline [2, 13, 20, 21]. However, previous studies were based on relatively small samples and short observation time.
In Nord-Trøndelag County, Norway, a large population based health study (the HUNT Study) combined with a registry of patients with dementia, provide data suitable for long follow-up time. The aim of this study was therefore to examine the association between hsCRP and AD in blood samples taken up to 15 years prior to the AD diagnosis amongst HUNT Study participants in Nord Trøndelag County over the age of 60 years.
Study population and data collection
The HUNT Study is a voluntary health survey offered to all residents in Nord-Trøndelag County (N~130,000). The region is approximately the size of Wales, rural, and located in central Norway. The HUNT Study consists of three population-based cohorts examining in total 125,000 residents during the span of three decades; HUNT 1 (1984-1986); HUNT 2 (1995-1997) and HUNT 3 (2006-2008). The HUNT Study has examined a large number of public health issues, like somatic and mental illnesses, quality of life, social factors, life style and other health determinants. The general methods for data collection were similar in all three HUNT surveys: several questionnaires, clinical measurements and collection of blood and urine samples. Participant’s age was obtained from the national population registry. History of myocardial infarction (MI), stroke, angina, diabetes mellitus (DM), smoking, and subjective health status were self-reported by participants. Clinical measurements were conducted in survey stations following standardized protocols. Pulse, systolic and diastolic blood pressure were measured three times using a Dinamap 845XT (Critikon) based on oscillometry. Body mass index (BMI) was based on height and weight measured with the participants wearing light clothes without shoes: height to the nearest centimeter and weight to the nearest half kilogram. Blood samples measuring non-fasting glucose, creatinine, triglycerides, and cholesterol used in the present study were collected at the health survey stations and transported to the biobank in well-described methods, that are published in detail previously [22, 23].
During HUNT 2 (n = 65,237) hsCRP measurement was measured in a subsample. For practical reasons, participants from four neighboring municipalities around the biochemical laboratory assaying hsCRP were selected randomly, and 9993 had their hsCRP measured (Fig. 1). The present study selected participants who had their hsCRP measured, returned the two main HUNT 2 questionnaires (n = 8766), and did not have prevalent dementia at time of survey participation (n = 8760). As hsCRP values can rise during active systemic infections or in acute inflammatory processes, we included only participants with hsCRP values less than 10 (n = 8391). Finally, we included only participants aged 60 and over (n = 2585) who had complete covariate data, which resulted in 2150 individuals who encompass the study sample. Non-fasting serum was stored at negative 80 degrees Celsius and measured two years after serum collection. The analysis were performed at a biomedical laboratory using the CRP (Latex) US (Hoffman-La Roche AG, Switzerland) standard assay for CRP analysis. Assay reproducibility was tested by the assay provider (Hitachi/Roche) and has run within [% coefficient of variation (CV) 0.43-1.34] and between days (%CV 2.51-5.70), in addition to running a method comparison (r = 0.996) .
The Health and Memory Study of Nord-Trøndelag (the HMS Study) collected retrospectively data on individuals with dementia from the two regional hospitals between 1995 and 2010. Additionally, residents in all nursing homes in the region were examined for dementia between 2010 and 2011and ascertained by clinicians. The data collection has been more extensively described previously . Briefly, two panels encompass the HMS study: a hospital and a nursing home panel. Ascertainment was uniform amongst panels and evaluated by clinicians confirming ICD-10 diagnostic criteria for AD, vascular dementia (VaD), and a mixture of these (mixed AD/VaD) based on clinical examination, patient and caregiver history and diagnostic imaging. Time of diagnosis was determined at assessment by clinicians and, if unknown, based on the initial documented examination date. The eleven-digit personal identification number given to each Norwegian resident linked the participants in HUNT 2 and individuals diagnosed with dementia in the HMS Study. Ninety-three HMS participants had hsCRP values less than ten and complete covariate data, of which 52 were diagnosed with AD; and are the focus of the present study. An additional 13 individuals were diagnosed with VaD, 12 mixed AD/VaD, and 16 with dementia of other causes.
Participant’s age was used as a continuous variable in analyses. Supplemental analyses were used to examine a significant interaction effect and created by dichotomizing participants >60 in equal groups <70.6 and ≥70.6. Level of education was categorized according to primary (seven years or less), secondary (seven to nine years), and upper secondary education (>ten years). The average of the second and third blood pressure measurement was used in analyses. Non-fasting glucose, cholesterol, triglycerides and creatinine were scored as continuous variables. The independent-samples t-test and Pearson’s chi squared were used to compare the means between groups for continuous and categorical variables, and Mann-Whitney (MW) for comparisons between cases and non-cases examining hsCRP levels and potential covariates. HsCRP was examined in analyses as a continuous variable. The values of hsCRP were positively skewed and log transformations were used in all analyses and were less skewed; but neither distributions were normal. We used binary logistic models to estimate odds ratios (OR) with 95% confidence intervals (CI) for the associations of hsCRP to the incidence of AD, all-cause dementia, and non-AD dementia. Four sets of logistic regression models were performed for each endpoint in a hierarchy. Effect modification was examined by testing the statistical significance for age x hsCRP and sex x hsCRP in multivariable models. Additional analyses were performed to examine whether time to ascertainment influenced the association by splitting the sample equally in two according to the number of years to diagnosis from baseline. All statistical analyses were performed using SPSS, version 24.
The participants in the present study had an average age of 73 years and were diagnosed with AD up to 15 years [mean 8.0 (±3.9)] following hsCRP measurement. The characteristics of the study sample are shown in Table 1. Mean hsCRP are shown in their original values. Levels of hsCRP were significantly lower in the AD group ≥70.6 than in the reference group. Except from the age and sex differences, there were no significant differences between the study groups regarding other biomarkers, education, and history of MI, angina, stroke or DM.
Multiple logistic regression analyses were performed for the total sample; and repeated separately for age groups 60- 70.5, and ages greater than or equal to 70.6 at the time of HUNT 2. The results for hsCRP in the total sample are shown in the upper part of Table 2 (1), in the sample aged 60-70.5 (2), and aged greater and equal to 70.6 in the lower portion (3). Results for total dementia, mixed AD/VaD, and VaD are presented in Table 3. There was no association between hsCRP and the risk of developing AD in the total sample, but there were significant age interactions in multivariate analyses. Additional analyses were performed with age dichotomized according to median age; and results of logistic regression analyses are shown in sections (2) and (3) of Table 2. In participants between 60 and 70.5, an adverse association was observed between hsCRP and AD. Conversely, in participants between 70.6 and 94, there was an inverse association between hsCRP and AD. Additional adjustment for all covariates did not change the finding.
Additional analyses were performed and presented in Table 2 (4-9) examining whether the number of years to AD onset from baseline influenced the association between hsCRP and AD. A similar adverse trend was observed amongst the sample diagnosed 0.4 to 7 years following hsCRP measurement. Amongst those 60-70.5, the adverse association between hsCRP and AD was attenuated and did not retain significance. An opposite trend was observed amongst those ≥70.6, where an inverse association was observed in participants diagnosed up to seven years later. The inverse association between hsCRP and AD amongst those diagnosed with AD seven-15 years later was attenuated and did not retain significance.
The main finding of our study was that hsCRP levels were adversely associated with participants aged between 60 and 70.5, and inversely associated with developing AD in participants aged ≥70.6. When applying multivariate models the findings were significant in individuals diagnosed only 0.4-7 years after the hsCRP was measured; and attenuated when AD was diagnosed more than seven years following hsCRP measurement.
Our findings support previous studies that report contrast findings when considering age. As in previous studies, participants in the younger age bracket (60-70.5) advocated that high hsCRP was associated with an increased risk of AD . In oldest participants, our findings support previous studies reporting an inverse association between hsCRP and AD [14, 15, 18, 19, 26,27,28,29,30].
Our study had a number of strengths in comparison with earlier studies, as a large number of subjects over the age of 60 had a follow-up time of up to 16 years. In addition, the prospective study design allowed for extensive control for numerous chronic conditions. Also, the utilized hsCRP assay has been shown to be a peripheral biomarker with high assertion. Our study should, however, be interpreted with some limitations. The HUNT Study participants are mostly Caucasian and the population is well educated, and results may not apply to all ethnicities or social demographics. The sample sizes in stratified analyses were relatively small. Although efforts were made to identify participants diagnosed with dementia in the region during 1995–2011 by performing hospital record searches and examining nursing home residents, we had no access to data from individuals with dementia who were under the care of their general practitioner, and these will appear as false-negatives in the data set. However, the proportion of false-negatives to true-negatives in the non-case group is quite low because the prevalence of dementia is, after all, low. Therefore, the contamination of the non-case group will not be substantial, and the effect estimates will be little more than inconsequential. Lastly, the prescription registry was not linked with the current study, and we cannot exclude that medication had an influence on hsCRP values, as it has been known that NSAIDs and lipid lowering medication such as statins reduce hsCRP values .
One challenge of the present study is to understand why hsCRP are in contrast when examining age of the participant during the years hsCRP is observed until the AD onset. It is questionable whether lower hsCRP values provides protection from AD, or if it is the result of the neuropathology in older at-risk individuals. A recent meta-analysis of CRP in persons with AD discussed whether CRP levels could be different in different stages of the disease trajectory. The authors speculated whether CRP is decreased in mild or moderate AD, and increased in the following severe stage . Dementia disorders are progressive and fatal disorders, as the blood samples obtained were an average of 8 years prior to diagnosis, it must be assumed that these were taken before AD developed or in very early stages. The results in this study appear to be more dependent on the age of the participant.
There have been a number of studies examining how immune responses can be affected by the pathophysiology of AD. Advances in neuroimmunology have shown that the molecular innate immune response is dysfunctional in AD . The body’s immune response in AD responds to an aggregation of amyloid-β (Aβ) peptides in the endoplasmic reticulum (ER) that causes stress and activation of the unfolded protein response (UPR) . UPR aims to alleviate stress and minor elevations of systemic inflammatory markers, reflects the presence of stressed cells. In circumstances of chronic or prolonged ER stress, sensors responsible for binding to misfolded proteins change from acting pro-protective to pro-apoptotic . It has been postulated that the molecular mechanisms involved in the innate immune response are disrupting UPR functioning and can be involved in the pathogenesis of AD . Although the precise molecular pathways of neuroinflammation remain unclear, a gene expression study found inflammatory changes in the aging brain regarded as age-dependent . Interestingly, the period between the sixth and seventh decade was observed to undergo robust gene expression changes.
It is known that clinical AD is preceded by decades of a prodromal phase. During this asymptomatic phase, systemic changes are known to be occurring. To examine whether our findings were influenced by ascertainment time, samples were split by the number of years participants developed AD following hsCRP measurement, see Table 2 (sections 4-9). There was a stronger association in participants who were diagnosed up to seven years later in comparison with those who were diagnosed seven to 15 years later. However, sample sizes in these stratified analyses were small and it is questionable whether the finding is a true association or the result of preclinical AD. Although, participants with dementia were ascertained in both nursing home and residential settings, it is perhaps speculative to say that nursing home participants were in a more severe stage than those at home, as there can be many other factors determining whether a Norwegian resident needs placement in nursing care. For example, those living in secluded areas, and often alone are demanding admission to a nursing home facility sooner than residents living at home with help from family and regardless of the stage severity. It is therefore difficult to distinguish strictly on this basis. Therefore, we examined stage severity using years to onset. Since the hsCRP marker was taken an approximately 8 years prior to diagnosis, it is most likely these participants were not exhibiting cognitive decline or at most, mild cognitive impairment.
Finally, low-grade inflammation is defined as being a state where the body is constantly under very mild chronic inflammation but not to the extent of acute inflammation. Minor elevation in inflammatory markers are measured in blood with inflammatory markers, such as hsCRP. Defining a precise cut-off between these two states is difficult, but many previous studies define a hsCRP under 10 with low-grade inflammation; and values above this as clinically significant inflammatory states . The American Heart Association have suggested that cut points of hsCRP below 1 mg/l, between 1 and 3 mg/l, and greater than 3 mg/l can be used to find those at lower, average, and high relative risk for CVD events . Replication of our data will strengthen the existing evidence whether similar cut points of hsCRP, in addition to a panel of other inflammatory markers, such as interleukins, should be considered clinically relevant when monitoring patients at risk for dementia.
Our study is in line with previous studies indicating a shift in the association between hsCRP and AD by age: in adults (60-70.5 years) there is an adverse association, while in seniors (>70.6 years) there is an inverse association. Regardless that the nature of the association remains unclear, our data and data from preclinical and clinical studies have established the immune system-mediated actions contribute and drive AD pathogenesis . Continued research in persons at risk is needed to advance the role inflammation has in AD. If our findings can be replicated, future intervention studies should assess whether medical treatment of low-grade inflammation will reduce incidence of AD. More studies are needed to further examine why a more active peripheral immune response may have a protective role in individuals ≥70 years.
Body mass index
C reactive protein
Coefficient of variation
Health and Memory Study of Nord-Trøndelag (1995–2010)
High specificity C reactive protein
- HUNT 1:
Helse Undersøkelse Nord-Trøndelag (1984–1986)
- HUNT 2:
Helse Undersøkelse Nord-Trøndelag (1995–1997)
- HUNT 3:
Helse Undersøkelse Nord-Trøndelag (2006-2008)
International Classification of Diseases, Tenth Revision
Non steroidal anti inflammatory drugs
Type two diabetes mellitus
Unfolded protein response
Heppner FL, Ransohoff RM, Becher B. Immune attack: the role of inflammation in Alzheimer disease. Nat Rev Neurosci. 2015;16(6):358–72.
Engelhart MJ, et al. Inflammatory proteins in plasma and the risk of dementia: the rotterdam study. Arch Neurol. 2004;61(5):668–72.
Paine NJ, et al. Induced mild systemic inflammation is associated with impaired ability to improve cognitive task performance by practice. Psychophysiology. 2015;52(3):333–41.
VanItallie TB. Alzheimer's disease: innate immunity gone awry? Metabolism. 2017;69s:S41–s49.
Eklund CM. Proinflammatory cytokines in CRP baseline regulation. Adv Clin Chem. 2009;48:111–36.
Nosalski R, et al. Novel immune mechanisms in hypertension and cardiovascular risk. Curr Cardiovasc Risk Rep. 2017;11(4):12.
Devaraj S, Siegel D, Jialal I. Statin therapy in metabolic syndrome and hypertension post-JUPITER: what is the value of CRP? Curr Atheroscler Rep. 2011;13(1):31–42.
Matusik P, et al. Do we know enough about the immune pathogenesis of acute coronary syndromes to improve clinical practice? Thromb Haemost. 2012;108(3):443–56.
Kushner I, Rzewnicki D, Samols D. What does minor elevation of C-reactive protein signify? Am J Med. 2006;119(2):166. e17-28
Woloshin S, Schwartz LM. Distribution of C-reactive protein values in the United States. N Engl J Med. 2005;352(15):1611–3.
Heneka MT, et al. Neuroinflammation in Alzheimer's disease. Lancet Neurol. 2015;14(4):388–405.
Michaud M, et al. Proinflammatory cytokines, aging, and age-related diseases. J Am Med Dir Assoc. 2013;14(12):877–82.
Schmidt R, et al. Early inflammation and dementia: a 25-year follow-up of the Honolulu-Asia aging study. Ann Neurol. 2002;52(2):168–74.
van Himbergen TM, et al. Biomarkers for insulin resistance and inflammation and the risk for all-cause dementia and alzheimer disease: results from the Framingham heart study. Arch Neurol. 2012;69(5):594–600.
Silverman JM, et al. C-reactive protein and familial risk for dementia: a phenotype for successful cognitive aging. Neurology. 2012;79(11):1116–23.
Katsel P, Tan W, Haroutunian V. Gain in brain immunity in the oldest-old differentiates cognitively normal from demented individuals. PLoS One. 2009;4(10):e7642.
Berchtold NC, et al. Gene expression changes in the course of normal brain aging are sexually dimorphic. Proc Natl Acad Sci U S A. 2008;105(40):15605–10.
Locascio JJ, et al. Plasma amyloid beta-protein and C-reactive protein in relation to the rate of progression of Alzheimer disease. Arch Neurol. 2008;65(6):776–85.
Nilsson K, Gustafson L, Hultberg B. C-reactive protein level is decreased in patients with Alzheimer's disease and related to cognitive function and survival time. Clin Biochem. 2011;44(14-15):1205–8.
Yaffe K, et al. Inflammatory markers and cognition in well-functioning African-American and white elders. Neurology. 2003;61(1):76–80.
Teunissen CE, et al. Inflammation markers in relation to cognition in a healthy aging population. J Neuroimmunol. 2003;134(1-2):142–50.
Krokstad S, et al. Cohort profile: the HUNT study. International Journal of Epidemiology: Norway; 2012.
Holmen J, et al. The Nord-Trøndelag health study 1995-97 (HUNT 2): objectives, contents, methods and participation. Norsk Epidemiologi. 2003;13(1):19–32.
Laugsand LE, et al. Insomnia and high-sensitivity C-reactive protein: the HUNT study, Norway. Psychosom Med. 2012;74(5):543–53.
Bergh S, et al. Cohort Profile: The Health and Memory Study (HMS): a dementia cohort linked to the HUNT study in Norway. Int J Epidemiol. 2014;43(6):1759-68. doi:10.1093/ije/dyu007. Epub 2014 Feb 12.
Yarchoan M, et al. Association of plasma C-reactive protein levels with the diagnosis of Alzheimer's disease. J Neurol Sci. 2013;333(1-2):9–12.
O'Bryant SE, et al. Decreased C-reactive protein levels in Alzheimer disease. J Geriatr Psychiatry Neurol. 2010;23(1):49–53.
Sundelof J, et al. Systemic inflammation and the risk of Alzheimer’s disease and dementia: a prospective population-based study. J Alzheimers Dis. 2009;18(1):79–87.
Tan ZS, et al. Inflammatory markers and the risk of Alzheimer disease: the Framingham study. Neurology. 2007;68(22):1902–8.
Wichmann MA, et al. Long-term systemic inflammation and cognitive impairment in a population-based cohort. J Am Geriatr Soc. 2014;62(9):1683–91.
Asher J, Houston M. Statins and C-reactive protein levels. J Clin Hypertens (Greenwich). 2007;9(8):622–8.
Gong C, et al. A meta-analysis of C-reactive protein in patients with Alzheimer’s disease. Am J Alzheimers Dis Other Demen. 2016;31(3):194–200.
Rivest S. Regulation of innate immune responses in the brain. Nat Rev Immunol. 2009;9(6):429–39.
Bernales S, Soto MM, McCullagh E. Unfolded protein stress in the endoplasmic reticulum and mitochondria: a role in neurodegeneration. Front Aging Neurosci. 2012;4:5.
Jager R, et al. The unfolded protein response at the crossroads of cellular life and death during endoplasmic reticulum stress. Biol Cell. 2012;104(5):259–70.
Torres JL, Ridker PM. High sensitivity C-reactive protein in clinical practice. Am Heart Hosp J. 2003;1(3):207–11.
We thank the HUNT-HMS study participants and their caregivers.
ExtraStiftelsen and the Norwegian Health Association are the study’s funding sources, and they had no other role in the HUNT-HMS Study. The corresponding author had full access to all data in the study and had final responsibility for the decision to submit for publication. The HUNT Study is a collaborative effort of the Faculty of Medicine and Health Sciences at the Norwegian University of Science and Technology (NTNU), the Norwegian Institute of Public Health, the Nord-Trøndelag County Council, and the Central Norwegian Regional Health Authority. The HMS Study was funded by the Norwegian Institue of Public Health, the NTNU, Nord-Trøndelag Hospital Trust, and Innlandet Hospital Trust.
Availability of data and materials
The funding institutions are currently granted exclusive privilege to perform studies on data derived from the HMS Study. Future collaboration with researchers outside the current research group will be welcome later. Information about application procedure is available on request from email@example.com.
Ethics approval and consent to participate
Participation in the HUNT study was voluntary and written consent was obtained. Approvals for the present study pertaining identifying individuals diagnosed with dementia during the HMS Study were acquired by participant and when appropriate from next of kin. The Regional Committee for Medical and Health Research Ethics in Mid-Norway approved the data collection in the HUNT study and in the HMS study, as well as the present study.
Consent for publication
The authors declare that they have no competing interests.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Kristian Tambs Deceased 18 June 2017