Marked inter-individual differences have been demonstrated in cognitive function and age-related decline (Foster, 2006). Understanding what causes this variability and identifying individuals at-risk of cognitive problems, is of critical importance to our aging population and workforce. Chronic low-grade inflammation, has been identified as one potential contributor to differences in brain age and cognitive function in older adults (Fard and Stough, 2019; Gordleeva et al., 2020).

C-reactive protein (CRP) is an acute-phase protein that has a critical role in the human immune system, and correlates with proinflammatory cytokine levels (Gabay and Kushner, 1999). Both CRP and cytokines found in blood can cross the blood-brain barrier to affect brain function (Quan and Banks, 2007). Animal studies have consistently shown an effect of inflammation on cognitive function. For example, administering the inflammatory agent lipopolysaccharide (LPS) to mice has been shown to cause learning and memory impairment (Salmani et al., 2020). LPS-induced inflammation impairs hippocampal neurogenesis in rats (Ekdahl et al., 2003), providing a putative mechanism connecting inflammation with brain function.

Elevated serum levels of CRP are considered a marker of systemic inflammation, and have been linked to several age-related conditions in humans, including cardiovascular disease, diabetes mellitus, and cognitive impairment (Cesari et al., 2003; De Rekeneire et al., 2006; Michaud et al., 2013). Higher CRP levels are also associated with reduced working memory (Bettcher et al., 2012); poorer executive function and processing speed (Tegeler et al., 2016); greater declines in visuospatial function (Warren et al., 2018); and smaller hippocampal volume across the lifespan (Wang et al., 2022). However, not all findings are consistent; discrepant findings may reflect differences in study design and sample characteristics, such as whether or not the participants being studied have an ongoing disorder (Carlier et al., 2021).

To date, few studies have investigated the relationship between genetic risk for CRP and cognitive function in typical aging. Those that have, are mostly lacking in statistical power or have focused on a limited number of individual CRP SNPs (~3 SNPs; for a review see, Stacey et al., 2017). For instance, one study used an unweighted PRS-CRP calculated using 47 GWAS significant SNPs and found that, after adjusting for sociodemographic and clinical confounders, contrary to hypotheses, higher PRS-CRP (and serum CRP) levels were significantly associated with better performance on a visuospatial memory test but not with reaction times (Milton et al., 2021). Investigation of PRS-CRP associations with a more comprehensive battery of cognitive tests is therefore warranted. Ultimately, if an association were confirmed, a PRS-CRP could be useful in prospectively identifying individuals at risk of both inflammatory conditions and poorer performance in specific cognitive tasks in adulthood. This would allow for the design of interventions such as exercise, anti-inflammatories, or brain training, to compensate for this predisposition (Bamidis et al., 2014; Melnikov et al., 2021).

We aimed to investigate the relationship between genetic risk for CRP levels and cognitive function in a sample of 7,817 individuals without dementia aged >50 years from the PROTECT study. Using the most powerful GWAS to date (i.e., N = 418,642; Han et al., 2020) we identified a PRS-CRP that explained 9.61% of the variance in serum CRP levels in an independent cohort (N = 268, age range = 20–84 years, females = 52.6%). We then assessed associations between PRS-CRP and performance on two cognitive batteries assessing a range of cognitive domains. Given evidence that CRP levels are negatively associated with cognitive function (Bettcher et al., 2012; Tegeler et al., 2016; Warren et al., 2018; Wang et al., 2022), we hypothesized that lower genetically predicted CRP levels would be related to better performance on the cognitive tests. Older individuals have the highest levels of CRP due to “inflammaging” (i.e., the age-related increase in systemic chronic inflammation, Fard and Stough, 2019; Gordleeva et al., 2020), and demonstrate more inter-individual variation in cognitive performance (see, Sánchez-Izquierdo and Fernández-Ballesteros, 2021), therefore we also used a median split to explore possible age-dependent associations specific to the oldest adults in our sample. Our results shed new and unexpected insights into the relationship between CRP levels and cognition in older adults.

To date, few studies have investigated the shared genetic relationship between CRP levels and cognitive function. We extended current research in this area by using a more powerful PRS-CRP and more comprehensive cognitive battery applied to a large sample of older adults. We did not observe any significant associations in the whole sample, even at the nominal level. Stratifying our sample by median age, in individuals aged 62 years and under, higher PRS-CRP was nominally associated with slower reaction times on the choice reaction task and lower accuracy on the digit vigilance task. Considering individuals aged over 62, higher PRS-CRP was nominally associated with better verbal reasoning and faster reaction times on the choice reaction task. Higher PRS-CRP was also significantly associated with better self-ordered search score in this older group (surviving correction for multiple testing).

Spatial working memory tasks, like the self-ordered search, are known to depend on the hippocampus (Spellman et al., 2015). Increased permeability of the blood-brain barrier with age may mean that CRP and inflammatory cytokines are more readily able to cross the blood-brain barrier to affect brain function (Erdö et al., 2016), and neurogenesis in the hippocampus (Ekdahl et al., 2003). This may explain why we observed effects in the older group only, albeit in the opposite direction to what we had hypothesized. That said, we did not find significant associations for all cognitive outcomes known to involve the hippocampus (i.e., the delayed visual recognition task).

When stratifying by median age we found some nominal effects in opposite directions (e.g., reaction time on the choice response task), which may mean that effects were masked when considering the sample overall. Our findings suggest that higher genetically predicted CRP levels may have positive effects in older individuals (>62 years) in specific aspects of cognition. This is in line with previous studies in adults over the age of 75 that found that increased CRP predicts better memory performance and decreased risk for cognitive decline (Silverman et al., 2009; Lima et al., 2014). An inverse relationship between CRP levels and onset of Alzheimer's disease has also been observed, whereby in adults >70.6 years, higher CRP was protective, but in adults aged 60–70.5 years it was adverse (e.g., Gabin et al., 2018). Perhaps relatedly, brain gene-expression studies have shown prominent changes in gene expression during the sixth to seventh decades, suggesting that this period is a critical transition point in brain aging (Berchtold et al., 2008) and that inflammatory genes are upregulated in brains of cognitively healthy older adults, but are down-regulated in dementia (Katsel et al., 2009). Others have suggested that a “protected survivor model” explains inverse age-dependent associations between CRP and cognitive aging, whereby the association of the risk factor with survival does not change within an individual, the association in the surviving population changes as its age increases due to differential mortality (e.g., Silverman and Schmeidler, 2018). Nonetheless, more research is needed to determine why higher peripheral levels of inflammation may have a protective role in healthy individuals ≥70 years. Our results and those of other studies highlight the importance of carefully considering the age of the sample in studies of (measured and genetically predicted) CRP and cognition in older adults.

The only previous study to our knowledge using a PRS-CRP to investigate the association between CRP levels and cognitive function in middle-older age adults (i.e., Milton et al., 2021), found a positive association between PRS-CRP and performance on the pairs matching test (visuospatial memory); though effect sizes were small. In this study, we did not find any associations between PRS-CRP and performance on the paired associates learning task (visuospatial memory). However, we did observe a statistically significant positive association between PRS-CRP and self-ordered search task performance (spatial working memory and executive function) in individuals aged over 62 years. Notably, both our results and those of Milton et al. suggest that higher genetically predicted CRP levels may be protective for spatial memory, contrary to hypotheses based on non-human animal work. Further, Milton et al. (2021) did not observe any associations between PRS-CRP and the reaction time test (processing speed). While we did not observe any associations between PRS-CRP and reaction times on the simple reaction time or digit vigilance task, we did observe a nominally significant association for reaction times on the choice reaction time task in analyses split by median age, with effects in opposite directions.

Together, our results suggest either the absence of a shared genetic relationship or a weak age-dependent and domain-specific genetic relationship between CRP and cognitive function in adults aged ≥50 years of European ancestry. Given the small effect sizes and limited significance of observed results, environmental factors experienced during the life course may better explain previous associations between CRP levels and cognitive function in middle-older aged adults than genetic factors. Indeed, prior infection has been linked not only to CRP levels but also to reduced cognitive function (e.g., Stebbins et al., 2021). Additionally, aside from pathogens, other more distal factors, such as pollution and traumatic events, have been linked to both elevated CRP levels and increased cognitive decline (Olvera Alvarez et al., 2018). It is also possible that higher peripheral inflammation negatively and preferentially impacts individuals already presenting with disease symptoms, or those at high risk of neuropsychiatric disease. For instance, the behavioral effects of LPS administration in rodents are modified by environment (e.g., stress; Barnum et al., 2012) and strain (e.g., Painsipp et al., 2008).

This current study had several limitations. Firstly, as is the case with many cognitive typical aging studies (see, Murman, 2015), the sample included an over-representation of women and those of a higher education level. While we adjusted for these variables in analyses, some caution is needed before generalizing findings to the wider population. The requirement for participants to have access to a computer and the internet could have further impacted the representativeness of the sample, particularly at older ages, where computer and internet use is less common (Niehaves and Plattfaut, 2014). Our results lack generalisability beyond European ancestry populations, namely because the GWAS of CRP levels from which the PRS was derived was based on individuals of European ancestry. Additionally, the ancestral composition of the cohort meant that we would have been underpowered to detect effects in other ancestral groups. Further research in more representative and diverse samples is therefore needed. Future research studies should collect blood for CRP assessment alongside biological material for genotyping, to ensure the effects of current CRP and genetic levels of CRP are accounted for when investigating the relationship between CRP and cognitive function. This would also facilitate drawing meaningful conclusions about environmental factors that may help to explain the observed phenotypic relationship between cognitive function and CRP levels. Mendelian randomization in larger samples of older individuals could also be used to confirm our mostly null findings.

Our results suggest either the absence of a shared genetic relationship or a weak positive relationship between CRP and working memory function in European adults aged ≥50 years. In light of prior research, the absence of a genetic relationship suggests environmental factors experienced during the life course might be affecting both CRP and cognitive measures, explaining previously reported associations.

The data analyzed in this study is subject to the following licenses/restrictions: This study was conducted using secondary data collected as part of the UK version of the PROTECT ongoing study. PROTECT data are available to investigators outside the PROTECT team after request and approval by the PROTECT Steering Committee. Requests to access these datasets should be directed to protect.data@exeter.ac.uk.

The studies involving human participants were reviewed and approved by London Bridge National Research Ethics Committee (reference: 13/LO/1578). The patients/participants provided their written informed consent to participate in this study.

AC, CB, DA, AH, and BC were involved in the design, funding, sample collection, and/or assessments within the ongoing PROTECT study. RA was involved in processing the genotype data and quality control. DD and MM were involved in designing the current work and advising on analysis strategies. AP and TP were involved in designing, analyzing, and writing the manuscript. All authors provided edits and approved the manuscript prior to submission.