Inactive bowel movement and stroke are associated with increased risks of mild cognitive impairment among community-living Singapore elderly

Introduction

Dementia, which occurs mainly in people older than 65 years, is one of the greatest global challenges for health and social care in the 21^st century [1]. Fifty million people worldwide lived with dementia in 2018, and the total number may triple to 152 million by 2050 [2]. The annual global cost of dementia treatment and care has reached one trillion US dollars and will double by 2030 [2]. Unfortunately, there is no effective disease-modifying pharmaceutical treatment available to cure or prevent dementia. Mild cognitive impairment (MCI) is a transitional state between the cognitive changes of normal aging and dementia [3], with a prevalence varying from 4% to 19% among people aged 60 and above [1]. MCI is also regarded as a preclinical phase of dementia, because around 22% of MCI-affected people would progress to dementia within 3 to 10 years upon diagnosis of MCI, while only 3% of non-MCI people develop dementia [4]. This pre-disease state of MCI opens an invaluable therapeutic time window to intervene or prevent further deterioration into dementia.

It was estimated that around one third of dementia burdens could be prevented theoretically by controlling modifiable risk factors [1]. Preventive countermeasures are composed of two major categories: management of medical comorbidities and modifications of lifestyles. First, high-risk medical comorbidities (e.g. midlife hypertension, obesity and hearing loss; late-life depression and diabetes) account for 17% of risks to develop dementia. Active treatments of these medical comorbidities were recommended to control MCI and prevent deterioration into dementia. Second, adopting healthy lifestyles (e.g. active engagements of social and physical activities, cognitive stimulation activities and smoking cessation) may control another 10% risks. Unfortunately, dietary factors were not included in the model, and their contributions are not calculated [1], although adherence to a Mediterranean diet was suggested to be beneficial for controlling vascular and metabolic risk factors. Previous studies have reported that adequate or excess intakes of fruit and/or vegetables were associated with a reduced risk of cognitive impairments [5]. Numerous preclinical studies have found that polyphenolic antioxidants found in fruits and vegetables could effectively decrease and even block neuronal death and thereby reduce the occurrence of dementia, especially in cell studies and in animal models [6].

Another close but neglected connection is bowel movement, which is affected by individuals’ dietary habits and the stool contents. Inactive bowel movement and constipation are common gastrointestinal symptoms (e.g., stool hardness, a prolonged and difficult passage of stool, or a feeling of unfinished evacuation) [7]. It has been proposed that gut-brain axis dysfunction may play a role in neurological diseases, strengthening the importance of gut flora in a biochemical signaling cascade between the gut and central nervous system. Constipation could precede clinical motor symptoms, particularly Parkinson’s disease (PD), by many years [8]. Based on a large-scale prospective cohort study (the Honolulu-Asia Aging Study) with 12 years’ follow-up, Abbott and colleagues made a landmark finding that infrequent bowel movement was correlated with an increased risk of PD [9]. A case-control study by Savica et al. using medical records documentation in the Rochester Epidemiology Project later reiterated the positive association between constipation (which was recorded 20 years or even more before the onset of motor impairments) and an increased risk of PD development. However, it is still unclear whether constipation may affect cognitive functions apart from motor neuron diseases. A few studies [10] have pointed out that patients affected by dementia with Lewy bodies had a higher prevalence of constipation. The authors suggested including examinations of constipation together with other non-cognitive symptoms, e.g., hyposmia and sleep behavior disorders, for the prodromal diagnosis of dementia.

In the present study, we aimed to dissect the associations of bowel movement frequencies and dietary habits with clinical-diagnosed MCI in a population-based cross-sectional study [the Diet and Healthy Aging (DaHA) study] in Singapore. For this observational study, we applied multiple logistic regression analyses to determine the adjusted odds ratios (AOR) of bowel movement on MCI occurrence. A Bayesian network approach that integrates causal diagram model and probabilistic model was then applied to infer the causal relationships between exposure variables (bowel movement frequencies and other possible influencers) and outcome variable (MCI occurrence). Qualitatively, the conceptual framework of causal association and the hypothesized causal-association paths among exposure variables were detected via a direct acyclic graph (DAG). Quantitatively, the magnitudes of causal association were inferred via the conditional probability table of the outcome variable under the combinations of direct factors. Moreover, the predictive reasoning forwarded from exposure variables and the importance scores of exposure variables were determined with a Bayesian network. This methodology has been conducted successfully in the field of neurological diseases. Kuipers and colleagues inferred the causal links between psychotic and neurotic symptoms in the British general population with directed acyclic graphs in a dynamic Bayesian network approach [11]. Rembach et al. employed the same approach to reveal the complex biological interactions specific to Alzheimer’s disease (AD) and assessed the biological connections among AD-associated serum protein biomarkers [12].

Results

Discussion

To the best of our knowledge, the present study was the first to examine the association between bowel movement and MCI. We found that active bowel movement, fruit consumption, and years of schooling were associated with decreased risk of MCI, while medical comorbidities including hypertension, stroke, and cataracts/glaucoma were unfavorable risk factors. The Bayesian network model further revealed that five hypothesized casual-association paths were important for MCI occurrence. Among them, inactive bowel movement and stroke were directly connected with MCI occurrence.

Gut-brain axis dysfunction has been hypothesized to be an essential factor in neurological diseases. Alterations of gut microbiota may mechanically promote neurological inflammation, insulin resistance [7, 13], and consequent neurological pathogenesis [14]. As a common digestive symptom, constipation (or inactive bowel movement) was well documented to increase the risk of motor neurological diseases. In addition to some case-control studies, the prospective Honolulu-Asia Aging Study [9] convincingly demonstrated that people with inactive bowel movement had increased risk for developing PD. However, there is a scarcity of solid epidemiological evidence associating bowel movement (or constipation) with cognitive impairments. Some observational studies have suggested a possible role of bowel movement in MCI or even dementia. On one hand, a few postmortem reports observed higher occurrences of dementia-related pathological changes (e.g., the presence of Lewy bodies and a decrease of substantia nigra neurons) in the brains of those who had been affected by constipation [15, 16]. On the other hand, some studies also reported that about 20% of dementia patients had constipation symptoms [17], which is consistent with findings of the present study that 26.7% MCI-affected elders had inactive bowel movement (Table 2). MCI-affected patients also observed gut microbiota alterations [18]. More importantly, these constipation symptoms can precede the onset of memory impairment by years.

The current study might support the gut-brain axis hypothesis by demonstrating that bowel movement was independently associated with MCI occurrence after adjusting for other influencing factors (Table 3). Furthermore, both bowel movement and stroke were directly associated with MCI in the causal diagram model of Bayesian network. The combination of inactive bowel movement and stroke increased the conditional probability to 60.00%, which was higher than inactive bowel movement alone or stroke alone (Table 4). Very recently, GV-971 (a sodium oligomannate) was found to be effective in relieving neuroinflammation and preventing cognitive impairments in a mouse AD model [19]. The authors concluded that the inhibitory actions of GV-971 on gut dysbiosis were responsible for the observed neuroprotective effects. Some promising results of a phase-III randomized clinical trial using GV-971 to treat mild dementia patients were presented in a conference, but the full results have not been published. It is worth noting that GV-971 was given conditional approval to treat patients with dementia by the China Food and Drug Administration in November 2019. Whether this drug could prove long-term efficacy in multi-regional randomized clinical trials and whether the proposed gut-brain axis dysfunction theory is valid warrant further studies.

The Bayesian network model indicated that education and fruit consumption were two important upstream factors influencing MCI, and they have the highest importance scores at 0.26 and 0.25 respectively (Figure 1). These two factors formed one hypothesized causal path directing from education to fruit consumption and then to MCI. This result is in line with a previous study observing that elderly with high education levels had higher fruit intake [20]. It is well established that childhood education enables individuals to increase cognitive reserve and even maintain cognitive function when they grow old. Education beyond primary school is regarded as the second important modifiable beneficial factor for dementia [1]. In the present study, around 42% of resident elders had no more than 5 years of education. Our results consistently showed that education (years of schooling) was negatively associated with MCI occurrence (Table 3). On the other hand, frequent consumption of fruit (≥4 days per week) was also negatively associated with MCI occurrence (AOR = 0.516, 95% = CI 0.330-0.808, P = 0.004) in our study (Table 3). Because education and fruit consumption belong to the same hypothesized causal path, the combination of poor education (≤ 5 years of schooling) and infrequent fruit consumption (≤3 times per week) reached a conditional probability at 32.86% (Supplementary Table 2), which is far less than the combination of the two direct factors (inactive bowel movement and stroke) with a conditional probability at 60.0% (Table 4).

Besides bowel movement and education via fruit consumption, the other three hypothesized causal-association paths directing to MCI were stroke, hypertension via stroke, and hypertension via cataracts/glaucoma, according to the Bayesian network model of causal inference (Figure 1). These medical comorbidities were statistically associated with MCI occurrence in the multiple logistic regression analyses (Table 3). Stroke and hypertension are well-known risk factors for MCI and dementia [1, 21, 22]. Active treatment of hypertension, however, significantly reduced the risk of stroke and MCI [23]. In addition, hypertension increases the risk of cataracts and/or glaucoma in population studies and rat experimental models [24]. Cataracts and glaucoma caused visual impairment, which usually affected health-related lifestyles (such as participations in social, physical, and cognitive-stimulating activities). However, these health-related lifestyles were well-recognized beneficial factors of MCI [1, 14]. A cross-sectional study in Japan showed that the elderly with mild visual impairments had more than 2 times higher risk of developing dementia [25]. In contrast, cataract surgery may help visual-impaired patients recover eyesight and lower significantly risks of MCI [25].

Based on the Bayesian network model of causal inference (Figure 1), there are three upstream influencers of bowel movement, namely fruit consumption, hypertension, and cataract/glaucoma. Because these three factors converge on the same node of bowel movement, no direct causal path could be inferred. Commonly consumed fruits could promote a healthy microbiota ecosystem by the actions of their fiber and polyphenolics. This conclusion was supported by a recent randomized controlled trial that consumption of fresh mango (a common tropical fruit in Singapore) significantly improved constipation status (frequency of bowel movements, stool consistency and shape, and duration of constipation) in chronically constipated adults [26]. Although bowel movement was associated with fruit consumption (Supplementary Table 1 and Figure 1), the multivariate logistic regression study confirmed their independent associations to MCI occurrence (Table 3). Here we did not find an association of green vegetable consumption with MCI occurrence, probably because the Southeast Asian residents maintain a good dietary habit of consuming green vegetables regularly. In our study, up to 89.2% of the elders ate green vegetables regularly no less than 4 days per week. Thus, it may be advisable for elders to consume fruits more frequently or daily to promote bowel movement and further prevent cognitive impairments.

We found that the elders with inactive bowel movement (≤3 days per week) had a higher proportion (57.4%) of hypertension comorbidity, compared to the peers with active bowel movement (46.9%, P=0.024, Supplementary Table 1). Whether there is a biological causal association directing from hypertension to bowel movement is yet unclear. However, recent studies suggested that the first-line antihypertensive drugs (e.g. diuretics, beta blockers, calcium channel blockers, and angiotensin-converting enzyme/angiotensin receptor blockers) were associated with an increased likelihood of constipation [27, 28].

Finally, a hypothesized connection between cataracts/glaucoma and inactive bowel movement was also suggested by the Bayesian network model of causal inference (Figure 1). The biological connection between these two factors has not been reported in the literature. We speculate that the hypothesized connection might be attributable to behavior or dietary changes among those elders affected by cataracts/glaucoma. Due to compromised vision, the affected elders may experience physical inactivity, which could lead to slower bowel movement. In addition, they may choose to prepare simple dish or have fast food with insufficient fiber consumption, another cause of inactive bowel movement. But these behavior/dietary connections warrant further examination in future cohort studies.

Conclusion

In summary, this study demonstrated that active bowel movement and fruit consumption were significantly associated with decreased risks of MCI in the elderly; while medical comorbidities including stroke, hypertension and cataract/glaucoma were positively associated with MCI. The Bayesian network model of causal inference further revealed five hypothesized causal-association paths leading to MCI, namely stroke, bowel movement, years of schooling via fruit consumption, hypertension via stroke and hypertension via cataracts/glaucoma. Particularly, the combination of stroke and inactive bowel movement reached a maximum conditional probability of 60.00%. Therefore, it would be advisable to decrease risks of MCI by adopting a healthy fruit consumption habit, improving bowel movement and by well management of stroke and hypertension.

Materials and Methods

Supplementary Materials

Supplementary Tables