Background
Dementia leads to disability and dependency among older people worldwide and thereby has enormous physical, psychological, social and economic impact on patients, caregivers, families and society [
1]. Alzheimer’s disease (AD) is the most common form of dementia accounting for 60–70% of the cases [
1]. In AD, pathologic protein aggregates and neuroinflammation mediated by microglia cells are involved in the pathogenesis, however, the exact mechanism is still unclear [
2]. Microglia maturation and function is critically dependent on short-chain fatty acids produced by the gut microbiome and therefore highlights the microbiome as a potential diagnostic and therapeutic target in dementia [
3]. During ageing, the gut microbiome decreases in diversity, loses beneficial taxa and facultative pathogens increase [
4]. Diet and the place of residence play an important role in the shaping of the microbiome [
5,
6]. Ageing is also associated with inflammation – often termed as “inflammageing”. Inflammation is further associated with an increase in gut permeability, mucosal inflammation and bacterial translocation [
2].
Since the main risk factor for developing dementia, especially AD, is ageing, it can be hypothesized that the gut-brain axis is a possible link between age and disease related dysbiosis and inflammation. Animal studies suggest that AD is associated with changes in the gut microbiome composition with a decrease in beneficial, anti-inflammatory genera [
7]. Furthermore, genetic alterations in amyloid genes can influence microbiome composition in mice, pointing towards a vicious cycle in AD development [
8]. Recently studies from the USA have identified a loss in species diversity and differences in bacterial composition in the stool of AD patients compared to matched controls [
9]. A study from Japan has also shown that gut microbiome composition is independently and strongly associated with dementia [
10]. Furthermore, it has been recently published that the microbiome of dementia patients causes a dysregulation of the anti-inflammatory P-glycoprotein pathway [
11].
So far factors that may influence the composition of the gut microbiome in patients with dementia have not been studied in detail. Potentially influencing factors could be malnutrition, which is common in dementia and associated with disease severity, [
12‐
14] or drug intake, since polypharmacy is a common problem in elderly persons and the impact of drugs on the microbiome has recently gained attention [
15‐
17].
We hypothesize that dementia is associated with dysbiosis, gut barrier dysfunction and inflammation and we aim to identify external factors influencing microbiome composition in dementia, such as nutrition and drug intake. To study this, we conducted a prospective controlled cohort study in patients with dementia and age matched controls.
Discussion
Our cross-sectional controlled pilot cohort study shows that dementia is associated with changes in microbiome composition including a reduction in bacteria known to produce short chain fatty acids (SCFA) and increased biomarkers of gut permeability and inflammation. Furthermore, we could show that both malnutrition and drug intake are factors associated with microbiome composition in dementia. This study therefore supports the concept of a disrupted gut-brain axis in dementia.
The concept of a disrupted gut-brain axis in dementia has recently emerged. Several animal studies show that induction of dysbiosis by antibiotics, irradiation or germ-free conditions negatively impact on cognitive function and plaque deposition as recently reviewed by Ticinesi et al. [
38] Recently, in patients with dementia a reduction in diversity of the microbiome has been described, however data on taxonomic microbiome composition are varying between different studies from different geographical locations [
9,
10,
39]. We describe altered beta diversity and distinct taxonomic changes in dementia in a European cohort. Alpha diversity data in dementia so far are conflicting, since a lower alpha diversity has been observed in the study from the USA [
9], whereas in the Japanese study a lower alpha diversity was observed in the control group [
10] but we found unchanged alpha diversity of the gut microbiome between European dementia patients and healthy age matched controls. Microbial diversity depends on many factors, especially in elderly, where the microbiome is likely to be less stable [
40]. Elderly people are more often exposed to microbial community-altering events, such as infections and concomitant antibiotic use, polypharmacy or hospital stays. Therefore, elderly controls may not be healthy in a strict sense and selection of the control cohorts may account for the different findings in different studies. Differences in analysis techniques [
41] and also in geographic location [
42] may be further factors that impact on diversity and composition of the gut microbiome.
When looking at taxonomic differences, abundance of
Eubacterium rectale, an uncultured
Lachnospiraceae sp. and
Lachnospiraceae NK4A136 group was lower in dementia patients compared to controls. LEfSe also identified the family
Lachospiraceae with its genus
Lachnospiraceae NK4A136 and several
Lachnospiraceae species to be associated with health.
Eubacterium rectale is a well-known butyrate producing bacterium [
43] and has already previously been associated with cognitive decline [
44]. Members of the
Lachnospiraceae family have been linked to obesity on the one hand and protection from colon cancer in humans on the other hand, likely due to the association of many species within the group with the production of butyric acid, a SCFA that is important for host epithelial cell growth and integrity [
45]. Mild dementia was also associated with another butyate producer –
F. prausnitzii [
46]. Increasing the number of butyrate producing bacteria in dementia may therefore be a promising therapeutic approach, since SCFA such as butyrate are critically involved in microglia maturation and function [
2,
3]. Data from animal and human pilot studies support this concept. A dietary intervention with bilberry anthocyanin extract was able to increase
Lachnospiraceae NK4A136 group abundance and improve gut barrier function in ageing rats [
47]. An exploratory pilot study in patients with dementia showed that a multispecies probiotic can increase the abundance of butyrate producing bacterial strains [
48]. We also found that the abundance of
C. clostridioforme and the genus
Eisenbergiella increased with increasing cognitive impairment.
C. clostridioforme has mainly been described as a human pathogen [
49] but has also been described to be associated with vegetarian diet [
50]. The genus
Eisenbergiella was recently found to be increased in long lived adults [
51]. Therefore these findings are difficult to interpret in the context of cognitive dysfunction. The family
Lactobacillaceae was differentially abundant in different stages of dementia, with the highest abundance in moderate dementia and a lower abundance in mild dementia and severe dementia. LEfSe also revealed the bacterium
L. amylovorus and the corresponding genus
Lactobacillus, the family
Lactobacillaceae and the order
Lactobacillales to be associated with moderate dementia. This finding is also difficult to interpret, since
Lactobacillus sp. in general were already more than 100 years ago associated with longevity by the Nobel prize winner Elie Metchnikow in 1907 [
52] and several studies using different
Lactobacillus sp. have been conducted with varying success in neurodegenerative diseases [
53].
Lactobacillus amylovorus has been described as a novel probiotic strain that is able to reduce ammonia levels and may therefore be associated with cognitive function [
54].
Our study also shows that dementia is not only associated with dysbiosis but also associated with markers of increased gut permeability (DAO) and markers of inflammation (sCD14). Ageing itself has been associated with an increase in gut permeability, mucosal inflammation and bacterial translocation – often termed as “inflammageing” [
2]. Increased calprotectin levels in stool as a sign of intestinal inflammation have been observed in a pilot study [
55]. Another study in dementia showed a decrease in previously elevated zonulin levels after probiotic treatment as a possible hint towards a causal link between dysbiosis and gut permeability in dementia [
48]. Although we did not find any differences in stool zonulin and calprotectin levels, we found an increase in DAO levels which has been proven to be a valuable serum biomarker of gut barrier dysfunction [
56‐
59]. Also, Ginko biloba, a commonly used phytotherapeutic drug in dementia, was able to reduce DAO levels in an animal model of alcoholic liver disease, indicating both the validity of DAO as a permeability biomarker and that gut hyperpermeability may be a modifiable and relevant therapeutic target [
60]. We furthermore found elevated sCD14 levels in dementia as a marker of endotoxemia and inflammation. Recent in vitro data suggest that the altered stool microbiome composition in dementia directly modulates intestinal epithelial homeostasis via the anti-inflammatory P-glycoprotein pathway [
11].
In our cohort, dementia patients, although not different regarding age and gender from our controls, received 3 times more prescription drugs. Although some of these drugs were only prescribed on demand, this finding is still striking. The known consequences of polypharmacy are among others, cognitive impairment, a higher risk of falls and non-compliance, but interventions to reduce polypharmacy are difficult [
61,
62]. Drug-microbiome interactions are increasingly recognized. A population based deep sequencing study revealed, that proton pump inhibitors (PPI) were associated with the most profound microbiome changes, followed by statins, antibiotics, laxatives and beta blockers [
17]. It has been shown experimentally that not only classic antimicrobials but also many other human-targeted drugs have an extensive impact on human gut bacteria [
15]. We recently showed that PPI are one of the main drivers of dysbiosis in liver cirrhosis [
63,
64]. We therefore assessed the association of prescription drugs with gut microbiome composition. As expected, effects on overall community structure (alpha and beta diversity) were small. Each drug class was associated with distinct associations throughout different taxonomic levels between users and non-users. PPI use was associated with higher abundance of oral bacteria in the stool and statins and antihypertensive use was associated with an increase in SCFA producing bacteria. Due to the small sample size, the results have to be interpreted with caution and can only serve as pilot data that need to be explored in larger cohorts. Additionally, we assessed the impact of drug intake on markers of gut permeability, bacterial translocation and inflammation. We found that PPI use was associated with increased intestinal inflammation. This has been previously described in the context of other diseases [
65‐
67] and we have recently linked dysbiosis, gut permeability and intestinal inflammation to adverse outcome in patients with liver cirrhosis who use PPI [
63]. Antihypertensive use was associated with slightly, but significantly elevated CRP levels, which is most likely due to the underlying disease and not to the drug itself, since arterial hypertension is associated with elevated CRP levels [
68] and therefore validates the relevance of our findings although sample size is small.
Malnutrition is common in dementia and nutrition care is an integral part of dementia care [
69]. Although all patients in our dementia cohort were treated according to nutritional support standards that include oral nutritional supplements in patients with MNA-SF < 9, MNA-SF and laboratory parameters showed that more than three quarter of the dementia patients in our study were malnourished. It is therefore impossible from this cross-sectional pilot study to distinguish if malnutrition or dementia are the starting point of dysbiosis. This could only be answered by longitudinal studies. In general, malnutrition has been associated with differences in microbiome composition, such as loss of bifidobacteria, however, most studies were performed in malnourished children and not in elderly people [
70].
In order to identify the drivers of dysbiosis we performed multivariate analyses with all variables and excluded variables that showed high collinearity to understand the driver of dysbiosis. We found that BMI and statin use were the strongest influencing factors, underpinning the notion that malnutrition and prescription drug use drive microbiome composition in dementia. Also network analysis supports the close association of these factors. However, due to the small sample size, the results of our multivariate analysis have to be interpreted with caution and can be seen as hypothesis generation only. The results need to be confirmed in larger studies.
Our study has some limitations: First, the single center approach and the sample size limit the generalizability of the data. A combination of all studies on gut microbiome in dementia would be desirable, however, due to the lack of standards in sequencing techniques this would not be technically feasible. Second, we could not perform the gold standard analysis of gut permeability – the differential sugar absorption test – because of the cognitive impairment of our patients, who were not able to follow the instructions of the test. We overcame this by using a panel of serum and stool markers that do not require compliance with test instructions for the participants. And third, this study only provides cross-sectional data and can therefore not answer any questions regarding causality or cause-effect relationship between cognitive dysfunction dysbiosis and malnutrition. A longitudinal study is in planning.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.