17372-87-1 br Acknowledgments This work was supported by grants

This work was supported by grants from the Portuguese Foundation for Science and Technology (J.R.G.: SFRH/BD/51677/2011, A.M.C.: SFRH/BPD/99613/2014 and PTDC/BIM-MEC/0651/2012) and FEDER/COMPETE (PEst-C/SAU/LA0001/2013-2014). From CNC, the authors thank Prof. João Nuno Moreira for providing the LysoTracker® and Prof. Ana Cristina Rego, Dr. John Jones, and Filipa Simões for their help regarding the production of Aβ fibrils.

Dementias have a long 17372-87-1 of preclinical development [1,2]. With increase in life expectancy of the aging population, it is all the more important to understand early changes in cognitive function that presage cognitive impairment. Systemic inflammatory markers have been investigated for this purpose in middle aged and older people and have been found to be associated with many cognitive abilities including episodic memory, executive function, and global cognition [3–12].
Two biomarkers are often used in studies of cognition and inflammation: fibrinogen and C-reactive protein (CRP). Epidemiologic, genetic, or pharmacologic studies on the relationship of fibrinogen to cognition suggest that higher levels of fibrinogen in the peripheral system are related to worse cognition [13–16]. In the Aspirin for Asymptomatic Atherosclerosis Trial of older people 17372-87-1 in Scotland (aged 50–80 years), higher levels of baseline fibrinogen and CRP were associated with worse cognition during follow-up in cross-sectional design [9]. Also in Scotland, the Edinburgh Artery Study (aged 55–74 years) found that higher levels of fibrinogen at baseline predicted cognition deficits 14 years later in cross-sectional design [8]. The baseline levels were also associated with cognitive decline in change-scores longitudinal design, where only cognition is repeatedly measured.
Epidemiologic studies of the link between CRP and cognition have found results that cover the negative to positive spectrum of associations. In the Rotterdam Study (aged 55–106 years), higher levels of CRP at baseline were associated with worse cognition in cross-sectional design, but the baseline CRP levels were not associated with decline in cognitive function in change-scores longitudinal design [11]. Across the Atlantic, the Health, Aging, and Body Composition study found that being in the highest tertile of CRP levels at baseline had the same odds as being in the lower tertiles when predicting 4 years cognitive decline in change-scores longitudinal design [4]. Weak evidence is also found in the British Whitehall II study (mean age 48, standard deviation [SD] 6 years) where levels of CRP at baseline were associated with worse reasoning and vocabulary functions in cross-sectional design but were not associated with cognitive decline in change-scores longitudinal design [6]. Remarkably in the older old (age ≥75 years), raised CRP is associated with better cognition [10,17], suggesting some genetic factors are responsible; see also [2]. An active inflammatory system may be found protective, delaying cognitive decline among the older old.
From a clinical perspective, this literature on peripheral inflammation has yet to bear fruit. Some are equivocal about the possibility [16], but others are less sanguine about the use of blood-based biomarkers in practice [11]. Such a view may be hasty given the weaknesses in the literature. First, it is unclear whether the positive relationship between inflammation and cognition applies to the younger old (50–75 years), a group of some importance given the well-known long duration of development of severe cognitive impairment. Second, many of these studies used one baseline measure of inflammation against repeated measures of cognition, effectively explaining change scores or decline as the dependent variable (change-scores longitudinal design). This poses two problems, one of which applies to any outcome (dilution bias), whereas the another applies specifically to cognition (practice effect). As is widely acknowledged in longitudinal studies, such change-scores design is susceptible to regression dilution bias [2–8,11,18]. It is possible that prospectively collected repeated measures of both cognition and biomarkers will modify the results mentioned.