Systemic inflammation and late-life cognitive ability

Abstract

Objectives – Cognitive ageing is an inevitable part of human life. Research from disciplines such as epidemiology, medicine and neuroscience implicate a wide range of determinants in the pathophysiological processes that lead to clinical symptoms of neurodegeneration. Markers of systemic inflammation are postulated to play an important role in mechanisms underlying a neuro-pathological cascade, either directly, through neuro-inflammatory processes, or through the mediating effect of diseases that are associated with cognitive deficits, such as cardiovascular disease and variation and disruption to cerebral blood flow. This may be particularly important in people with type 2 diabetes, where the increased prevalence of vascular events and glycaemic upset along with elevated levels of various circulating biomarkers, have been implicated in accelerated cognitive decline. Increasingly, evidence suggests a contribution of vascular disease state in the development of Alzheimer’s disease in which inflammation could be a significant factor. Determining the direction of association between individual markers of inflammation and altered cognitive performance is important in order to understand the possible role of inflammation in the development of cognitive decline and to inform the development of preventive clinical interventions. Therefore investigating these risk factors in relation to the trajectory of age related cognitive decline is crucial; in this respect, longitudinal evidence, detecting change in cognitive performance over a defined period of time, is most appropriate. To date, the majority of evidence is inconclusive, predominantly due to methodological obstacles embedded in the prospective design of cognitive ageing studies and in the investigation of a complex disease state, such as insufficient follow up period and restricted cognitive assessment. Since associations reported from modelling late life cognitive change in epidemiological studies may be the result of confounding variables, such as gender, vascular risk factors/disease, education attainment and social status, investigating the causal nature of inflammatory mediators in cognitive decline, has proved more problematic. Additionally, even a casual association may be due to ‘reverse causation’. One method of unravelling such associations is through the use of genetic association, where the exposure variable of interest (such as genetic variants affecting plasma biomarker levels) is modelled against the outcome, thereby overcoming some of the problems of confounding and reverse causation inherent in non-genetic epidemiological studies. Aim – The primary aim of this thesis was to test for associations of baseline measures of acute-phase proteins (fibrinogen and C-reactive protein) and central pro-inflammatory cytokines (interleukin – 6 and tumour necrosis –α) with four-year change and estimated life-time change in cognitive ability in older people with type 2 diabetes. The second aim was to explore the association between fibrinogen-related SNPs (SNPs shown previously to be associated with altered plasma fibrinogen levels) and cognitive ability in the general population. Methods –Data from the Edinburgh Type 2 Diabetes Study (the ET2DS), a prospective epidemiological study of older people with type 2 diabetes were available, including that collected at a baseline clinic (2006-07) on 1066 participants, mean age 67.9 years (SD 4.2). For the present study, follow up cognitive assessment was carried out after four-years (2010-11) at which cognitive data was collected on 828 survivors. Cognitive ability at both time points was assessed using the same, comprehensive, seven neuropsychological tests battery, including measures of fluid as well as crystallised intelligence (vocabulary test). Principal component analysis was conducted to derive a general cognitive factor ‘g’ and a general inflammatory factor, derived from individual cognitive scores and from baseline measures of four inflammatory markers (fibrinogen and C-reactive protein, interleukin – 6 and tumour necrosis –α), respectively. Genotype and cognitive data were collected from seven, well-established population-based cohorts with clearly defined sampling frames and data collection procedures. Five cohorts comprised of community-dwelling elderly people, living in central Scotland (AAA Trial, n = 2061, EAS, n = 534; ET2DS, n = 1045; LBC 21, n = 517; LBC 36, n=1005) and two large were cohorts based in England (ELSA; n = 5458; and Whitehall II; n = 3400). In total, genotype and cognitive data were available for 14033 participants, age range between 60 to 80 years. In all studies cognition was assessed on three cognitive domains: memory, executive functioning and information processing. Compatibility of cognitive data allowed for calculation of a general cognitive factor ‘g’ that was comparable between all cohorts. The instrument variables consisted of 61 fibrinogen-related polymorphisms within 13 different loci. These were identified through a detailed literature search as well as through search of relevant, genetic databases. Results – in the ET2DS, the age and sex-adjusted analyses revealed statistically significant associations between raised plasma inflammatory markers and poorer ‘g’ at follow-up; this was observed for all biomarkers, with the strongest associations detected for IL-6 and the general inflammation factor (p values <0.001). These findings persisted in linear regression models of baseline biomarker levels with four-year cognitive change as well as estimated life-time change – here the general inflammatory factor and plasma IL-6 levels were the strongest predictors. Adjustment for conventional vascular risk factors and cardiovascular disease attenuated the associations of cognitive decline with fibrinogen, CRP and TNF-α; associations were largely attenuated in analyses assessing IL-6 and the general inflammation factor and tended to remain statistically significant. Meta-analysis was conducted in order to explore associations between pre-selected fibrinogen-related SNPs and impairment in general cognitive ability as indexed by ‘g’. The analysis identified five plasma fibrinogen-related SNPs that were significantly associated with impaired ‘g’ at the nominal threshold level of p < 0.05. These were: rs2070016 (FGB gene); rs2070016 (FGA gene); rs1800497 (ANKK1 gene); rs4251961 (IL1RN gene) and rs1130864 (CRP gene). Discussion – the results of the ET2DS indicate that in an elderly diabetic population, there is a significant relationship between baseline levels of circulating inflammatory markers and four-year cognitive change as well as estimated life-time cognitive decline. These associations were generally independent of common cardiovascular risk factors and events, suggesting a possible pathway where cytokine-induced activation of glial cells may be responsible for the consequent neuro-inflammatory processes resulting in declined cognitive ability. The lack of some associations may be due to a relatively short follow up period. The main strength of the ET2DS was the availability of prospective cognitive data, the large sample size and the use of a comprehensive cognitive battery, including a vocabulary test for crystallised intelligence and thus calculation of estimated life-time cognitive change. Genetic association analysis indicated a significant association between five preselected SNPs each located within different genes (in general, genes associated with inflammation), and impaired general cognitive ability. This provides some support for a causal role of inflammation in age-related general cognitive impairment. One of the major strength was the use of a large dataset and the applied methodological approach. Meta-analysis was conducted on raw, prospectively generated data, allowing determination of the cognitive phenotype variable. The main outcomes of this thesis suggest that systemic inflammation may indeed be involved in aetiology of age-related cognitive decline, possibly via neuro-inflammation. Further epidemiological investigation should involve a measurement of biomarkers trajectories in modelling cognitive change. Use of a stronger genetic instrument for inflammatory biomarkers, modeled against cognitive decline rather than cognitive ability as in the current study could further advance knowledge of the bio-pathological mechanisms underlying age-related cognitive decline. Results could ultimately inform subsequent investigations in the form of a randomised control trial, testing an evidence-based anti-inflammatory clinical intervention in diabetic populations as well as the general populations.