With rapid population aging worldwide, cognitive disorders, such as dementia, pose a serious concern for public health and to the social and economic development of modern society (Decarli, 2003; Fox and Petersen, 2013). Dementia is a major cause of disability and for the institutionalization of older people. There is no effective disease-modifying treatment currently available (Norton et al., 2013; Livingston et al., 2020). A substantial amount of evidence shows that both genetic and lifestyle factors play a role in determining the individual risk of dementia and cognitive impairment. The ε4 allele of the apolipoprotein E (APOE ε4) gene is the strongest known genetic risk factor for dementia and cognitive impairment, and it may represent roughly half of the genetic risk for neuropathologically based cognitive diseases (Raber et al., 2004; Caselli et al., 2010; Lim et al., 2018). Nevertheless, the genetic risk of APOE ε4-related cognitive decline could be partly offset by favorably modifiable factors, such as higher education, more intake of protein-enriched food and choline, and maintaining vascular health (Ferrari et al., 2013; Sienski et al., 2021; Zhang et al., 2021). More specifically, established studies have suggested the protective effects of these modifiable factors, including engaging in leisure activities, were more pronounced in APOE ε4 carriers (Goodwin et al., 1983; Ott et al., 1998; Huang et al., 2005; Rovio et al., 2005; Kivipelto et al., 2008).

Leisure activities have generally been thought to improve cognitive health, per the “use it or lose it” hypothesis (Salthouse, 2006). It should be acknowledged that there are varied types of leisure activities, typically physical, social, and productive activities. The three leisure activities have varied pathways which converge within three major etiological hypotheses for dementia and other cognitive disorders: the cognitive reserve hypothesis, the vascular hypothesis, and the stress hypothesis (Fratiglioni et al., 2004). It is valuable to differentiate types of leisure activities when studying their effects on cognitive decline, especially for the people who carry an APOE ε4 allele, since it would provide useful information for designing effective intervention programs to prevent cognitive decline and other neurodegenerative diseases. In addition, the extant two studies with either relatively small sample size or short duration of follow-up reported inconsistent results, regarding the interaction effect of APOE ε4 with physical, social, and productive activities on cognitive health (Niti et al., 2008; Yang et al., 2015). We, therefore, used a nationwide, community-based longitudinal dataset with a relatively long duration of follow-up (about 6 years) to test the associations of APOE ε4 allele carriage and engagement in leisure activities with cognitive decline and to investigate whether there were interactive effects of APOE ε4 carriage with leisure activities. We also evaluated associations of the three leisure activity subtypes on cognitive decline. We hypothesized that there would be a significant statistical interaction where carriers of the APOE ε4 allele who engaged in leisure activities would have lower odds of cognitive decline compared to the noncarriers.

In this study, we found that engagement in leisure activities and APOE ε4 carriage had significant and opposite effects on cognitive health, consistent with previous studies. Among the three sub-types of leisure activities, our findings showed that the negative associations with cognitive decline were more pronounced in productive and social activities than in PAs. Our study was one of only a few that reported significant interactions between leisure activities and APOE ε4 on cognitive decline, indicating that practicing leisure activities, particularly productive and social activities, might partly offset the risk of APOE ε4-related cognitive decline.

There was consistent epidemiological evidence linking lack of participation in leisure activities with the development of dementia and decline of cognitive function (Wilson et al., 2007; Niti et al., 2008; Akbaraly et al., 2009). A mechanism hypothesized to underlie this linkage was that leisure activities exert protective effects on cognitive function by reducing stress (McEwen, 2002). Stress could cause corticosterone hypersecretion, which downregulates the hippocampal corticosteroid receptors, which in turn dampens the feedback inhibition of the adrenocortical axis leading to further hypersecretion, which eventually harms the hippocampal neurons (Sapolsky et al., 1986). Leisure activities could lead to positive emotional states, such as self-esteem, social competence, and adequate mood, all of which are buffers to stress (Fratiglioni et al., 2004). Findings from the current study provided evidence that higher scores of ILAs were associated with lower odds of cognitive decline, suggesting that integrated leisure activities could protect against cognitive decline.

Results from a sub-group analysis of productive activities, social activities, and PAs showed that among the three sub-types, productive activities have the most robust association with lower odds of cognitive decline, compared with social activities and PAs. Previous studies have shown that productive activities were associated with the reduced risk of dementia and cognitive decline. For example, in a French cohort, knitting, doing odd jobs, gardening, and traveling reduced the risk of dementia (Fabrigoule et al., 1995). In the Nun Study, low density of ideas and low levels of grammatical complexity in autobiographies written in early life were associated with low cognitive test scores in later life (Snowdon et al., 1996). In the Bronx Aging study, reading, playing board games, playing musical instruments, and dancing were associated with a lower risk of dementia (Scarmeas, 2003). It has been suggested that mental stimulation could selectively increase synaptogenesis (Churchill et al., 2002) while accumulating evidence concerning brain plasticity in adult life has shown the existence of angiogenesis, synaptogenesis, and neurogenesis (Fillit et al., 2002). Productive activity, which was more relevant and directly associated with mental stimulation in contrast to social activities and PAs, may create more dendritic branches and synaptic connections and alleviate the amyloid burden in the brain, which improves or preserves cognitive function (Fratiglioni et al., 2004). This was supported by evidence from studies showing that activities with higher cognitive demand have a more robust association with cognition (Hultsch et al., 1999) and may protect against cognitive decline (Katzman, 1995; Wang et al., 2006). Meanwhile, our findings also reinforced the arguments that social activities may affect cognitive health through common pathways by receiving emotional support from relatives and friends thus reducing depression and the adverse effects of stress and associated elevation of cortisol (Bassuk et al., 1999). Previous studies suggested that physical training may promote cerebral blood flow and angiogenesis, decreasing the accumulation of radical oxidative proteins, thus benefitting cognitive function (Kivipelto et al., 2001). We were unable to identify a significant beneficial effect of PA. Notably, the PAs among older people in this study were self-reported in contrast to formal, physical training. It is possible that the lack of professional advice and supervision for the PAs leads to limited power to protect against cognitive decline.

We did not detect the significant association between carrying APOE ε2 allele and slower cognitive decline, partly due to the small toward nonneuroprotective effect of APOE ε2 at an advanced age. In this study, we reported a significant association between APOE ε4 carriage and accelerated cognitive decline, consistent with the established literature. It is reported that APOE ε4 risk allele carriers have worse brain profiles, i.e., smaller brain volumes and poorer cognition, than noncarriers (Agosta et al., 2009; Honea et al., 2009; Schuff et al., 2009; Hafsteinsdottir et al., 2012). One study investigated the association of the APOE ε4 allele and leisure activity with brain tissue volume, in a population-based cohort of 4,303 non-demented individuals, aged 66–96 years. This study reported that APOE ε4 carriers had greater white matter hyperintensities and cerebrospinal fluid volumes than noncarriers but smaller gray matter volumes (Hafsteinsdottir et al., 2012). Individuals with lower leisure activity levels had greater white matter hyperintensities and cerebrospinal fluid volumes, smaller total brain parenchymal, white matter, and gray matter volumes than those with the highest levels of participation (Hafsteinsdottir et al., 2012). Therefore, the study concluded that carriers of the APOE ε4 risk allele have worse brain profiles than noncarriers. Participation in leisure activity is a factor associated with brain tissue volume and white matter hyperintensities load and, in turn, affects age-related cognitive decline (Hafsteinsdottir et al., 2012).

Significant interactions were found between leisure activities and APOE ε4 carriage in the current study, notably in the sub-type of productive activities. The protective effect of leisure activities on cognitive decline was more pronounced in APOE ε4 carriers. Based on the interpretation of brain tissue volume differences between APOE ε4 carriers and noncarriers, it has been postulated that individuals carrying the APOE ε4 allele may have less effective neural protection and repair mechanisms and maybe thus more dependent on lifestyle-related factors to protect them against cognitive decline and dementia (Mahley and Rall, 2000). Meanwhile, leisure activities, especially productive activities, could reduce rates of hippocampal atrophy (Valenzuela et al., 2008). Therefore, the protective effect of leisure activities, especially productive activities on cognitive decline, would be more pronounced in APOE ε4 carriers. Social activities could affect cognitive function by common pathways of reducing stress, and indeed there was a trend toward statistical significance regarding the interaction between social activities and APOE ε4 (P = 0.109), with the protective effect on cognitive decline more pronounced in APOE ε4 carriers. However, a significant interaction between PAs and APOE ε4 was not observed. Although some studies suggested that PA may modify the impact of APOE ε4 on Alzheimer's disease (AD) and cardiovascular disease (CVD) risk (Bos et al., 2018), however, the evidence was inconclusive; other longitudinal studies found no evidence for an interaction effect between PA and APOE ε4 in cognitive decline in older adults (Folley et al., 2019; Stringa et al., 2020) or found the association of PA with better cognitive function only in among APOE ε4 noncarriers (Fernández-Matarrubia et al., 2021). In the current study, our results were in accordance with no interaction. This may suggest that physical stimulations are less pronounced in APOE ε4 carriers than mental stimulations, nevertheless, the validity of self-reported physical exercise among older people in China should also be taken as a plausible explanation.

We also found that women were more sensitive to APOE ε4-by-leisure interaction (ILA) in contrast to men. Although the P-value of the three-way interaction was not significant, a plot of the interaction between APOE ε4 and ILA on cognitive decline showed that this interaction was principally driven by female participants. Established studies showed that lipid metabolism in women is likely to be more affected by APOE genotype than in men (Ungar et al., 2014; Sienski et al., 2021; Zhang et al., 2021). Meanwhile, studies of the hippocampus as a target of stress and sex hormones have revealed a considerable degree of structural plasticity and remodeling in the adult brain that differs between the sexes, with effects more marked in women (McEwen, 2002). In our study, changes in cognitive decline with aging, female APOE ε4 carriers were found to be the most sensitive, in contrast with female APOE ε4 noncarriers, male APOE ε4 carriers, and male APOE ε4 noncarriers.

Some aspects of the measurement of leisure activities and interpretation of the results require discussion. First, the activities in the sub-types may not be mutually exclusive. For example, doing housework was assigned to productive activities, but it may also be physically demanding. One of our criteria for separating leisure activities into sub-types was definition in established studies. Activities involved with more brain function would be assigned to productive activities, activities involved with other people would be assigned to social activities, and activities involved with more physical movements would be assigned to PAs. Second, self-reported leisure activities are context-specific due to social environment and culture, therefore, in this study, physical exercise among older people in China could differ from standard physical training in some developed countries. The differentiation of physical labor from leisure activities among older people in China is complex. Since there were still a large proportion of older people living in or having lived in rural areas, they often engaged in fieldwork, such as gardening for leisure, although they may no longer make their living in agriculture. On the other hand, physical exercises for some older people include physically involved activities that do not require specialized guidance, such as square dancing or walking. In those circumstances, the stimulation effect of physical exercise may be disused or not as effective as standard physical training.

Our study findings are of help to better define the target populations for future interventions and to better delineate preventive and therapeutic lifestyle strategies. The identification of lifestyle-gene interactions offers us an opportunity to develop leisure interventions, particularly productive and social activities that will mitigate the detrimental effects of genetic factors including APOE ε4 allele carriage. Leisure activities that stimulate cognitive function may be recommended as an intervention to modify the rate of age-associated cognitive decline in older people.

The main strengths of this study are the longitudinal design with a relatively long duration of follow-up (≥6 years) and the use of a nationwide sample of Chinese community-dwelling older adults (n = 3,017, with 588 APOE ε4 carriers). Another strength was that we classified leisure activities into productive leisure activities, social activities, and PAs for a better interpretation of the effect on cognitive decline. We also conducted a series of sensitive analyses to evaluate the statistical robustness of the results.

Our study has several limitations. First, our study could not distinguish potential reverse causation. This is because we could not “set” leisure activities but only observe them from the survey, therefore, the assumption that participation in leisure activities has a directional effect on cognitive decline could be reversed, e.g., cognitive decline and disease reduce the likelihood of participation in leisure activities. We controlled for potential confounding variables, such as diseases, which were risk factors for cognitive decline so that the association between baseline leisure activities and cognitive decline was not confounded by these factors. Second, we were unable to include more leisure components, such as listening to music, art creation, or writing due to the limitation of the questionnaire design of CLHLS. Third, we did not have a precise physician-based clinical diagnosis of cognitive decline. Therefore, future studies with more sophisticated designs and clinical endpoints are warranted to decipher the mechanisms of gene-by-lifestyle interactions between APOE ε4 and leisure activities.

Our study provided evidence that leisure activities, especially productive and social activities, can potentially offset the detrimental effect of APOE ε4 allele carriage on cognitive health. Further research with a clearer definition of the types and intensities of leisure activities is needed for a more precise understanding of this relationship. If substantiated, personalized lifestyle recommendations of having more productive leisure activities for the older people with the APOE ε4 allele may bring substantial benefits for maintenance of cognitive health.

The data analyzed in this study is subject to the following licenses/restrictions: The datasets during and/or analyzed during the current study available from the corresponding author on reasonable request. Requests to access these datasets should be directed to Yao Yao (yaoyao@nsd.pku.edu.cn).

YunZ, YY, and YiZ designed the research and directed its implication. YunZ, SF, and YY prepared and analyzed the data and drafted the manuscript. All co-authors revised the manuscript together.

This work was supported by the National Natural Science Foundation of China (81900357, 81903392, 81941021, and 72061137004), National Key R&D Program of China (2018YFC2000400), National Social Sciences Foundation of China (19CRK005), China Postdoctoral Science Foundation funded project (2019M663338 and 2019M650359), the U.S. National Institute of Aging of National Institute of Health (P01AG031719), the Duke/Duke-NUS Collaboration Pilot Project, and the Data for Better Health Project of Peking University-Master Kong.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.