Association Between Oral Bacteria and Alzheimer’s Disease: A Systematic Review and Meta-Analysis

Abstract

Background:

Pre-clinical evidence implicates oral bacteria in the pathogenesis of Alzheimer’s disease (AD), while clinical studies show diverse results.

Objective:

To comprehensively assess the association between oral bacteria and AD with clinical evidence.

Methods:

Studies investigating the association between oral bacteria and AD were identified through a systematic search of six databases PubMed, Embase, Cochrane Central Library, Scopus, ScienceDirect, and Web of Science. Methodological quality ratings of the included studies were performed. A best evidence synthesis was employed to integrate the results. When applicable, a meta-analysis was conducted using a random-effect model.

Results:

Of the 16 studies included, ten investigated periodontal pathobionts and six were microbiome-wide association studies. Samples from the brain, serum, and oral cavity were tested. We found over a ten-fold and six-fold increased risk of AD when there were oral bacteria (OR = 10.68 95% CI: 4.48–25.43; p < 0.00001, I² = 0%) and Porphyromonas gingivalis (OR = 6.84 95% CI: 2.70–17.31; p < 0.0001, I² = 0%) respectively in the brain. While AD patients exhibited lower alpha diversity of oral microbiota than healthy controls, the findings of bacterial communities were inconsistent among studies. The best evidence synthesis suggested a moderate level of evidence for an overall association between oral bacteria and AD and for oral bacteria being a risk factor for AD.

Conclusion:

Current evidence moderately supports the association between oral bacteria and AD, while the association was strong when oral bacteria were detectable in the brain. Further evidence is needed to clarify the interrelationship between both individual species and bacterial communities and the development of AD.

Keywords

Alzheimer’s disease cognitive impairment dementia microbiota periodontitis

INTRODUCTION

Alzheimer’s disease (AD) is a progressive neurodegenerative disease pathologically described by extracellular amyloid-β (Aβ) plaques and intracellular neurofibrillary tangles (NFTs) in the brain. As the most prevalent cause of dementia, AD is responsible for between 60% to 80% of total dementia cases [1]. Approximately 47 million people are living with this ultimately fatal disease or related dementia worldwide, and this figure is expected to triple by 2050 [2], imposing huge burdens on individuals, families, and the wider society.

AD has two forms: the inherited or familial form and the sporadic form [3]. Familial AD is determined genetically and accounts for only 5% of all AD patients, while sporadic AD constitutes the vast majority of cases and has a complex and multifactorial etiology in which infection is implicated [4]. The infectious etiology hypothesis of AD, first presented by Dr. Oskar Fisher over a hundred years ago has garnered more and more attention over the last three decades. Various infectious agents have been detected in postmortem AD brains and some proposed to contribute to cognitive decline, dementia, and AD [5]. Periodontitis is currently regarded as an inflammatory disease initiated by bacteria. Noteworthy, periodontitis is linked with a higher risk of AD by a number of epidemiological studies [6 –8]. In a retrospective study, this risk increased up to 1.7 times after 10-year exposure to chronic periodontitis [9]. The emerging link between this disease and AD has prompted researchers to look deeper into the role periodontal bacteria may play in the pathogenesis of AD. To date, exposure to certain periodontal pathobionts such as Porphyromonas gingivalis, Treponema denticola, and Aggregatibacter actinomycetemcomitans has been demonstrated to induce AD-like changes in animal models or neural cell cultures through multiple mechanisms, such as activating microglia, manipulating the complement system, insulting cerebrovasculature, enhancing the expression of pro-inflammatory factors and Aβ levels [10 –13]. The accumulating pre-clinical evidence is suggesting a role for oral bacteria in the pathogenesis of AD. By contrast, methodological variations and confounding factors often diversify the results and lead to inconsistencies across clinical studies, which obscures the relationship between oral bacteria and AD.

The oral cavity houses the second most diverse microbiome after the gut in the human body. To date, over 700 bacterial species have been detected intraorally [14]. Their interactions within microbial communities and with various host factors are a major driver of symbiosis or dysbiosis in the oral ecosystem, which can be accompanied by whole-scale changes of oral microbiota that impact both oral and general health [15]. Profiles of oral microbiota, in turn, are altered in a plethora of oral and systemic disorders such as poor oral health, diabetes, cancer, autoimmune diseases, and cognitive disorders [16]. Unravelling the changes of oral microbiota in these conditions is becoming critical for its diagnostic and therapeutic potential. Although this hidden information has been mined in the context of AD by a few studies, no consensus is reached. Moreover, hitherto research on the relationship between oral bacteria and AD mainly focus on specific periodontal pathobionts rather than the whole microbiome. Here we performed a systematic review and meta-analysis to qualitatively and quantitatively assess all of the published clinical data on the association between oral bacteria and AD. This work aims to gain a better comprehension of the extant evidence on how AD is related to both individual species and the microbial landscape in the human oral cavity, which may provide useful perspectives for future studies before any step toward microbe-oriented AD prevention, diagnostics or management can be justified.

MATERIALS AND METHODS

This systematic review was conducted following the Preferred Reporting Items for Systematic Reviews and Meta-Analysis guidelines.

Search strategy

We searched PubMed, Embase, the Cochrane Central Register of Controlled Trials, Web of Science, ScienceDirect, and Scopus databases for pertinent publications from the first available date to May 2022. The following search terms were used: ("Periodontal bacteria” OR “Oral bacteria” OR “Porphyromonas gingivalis” OR “Treponema denticola” OR “Tannerella forsythia” OR “Aggregatibacter actinomycetemcomitans” OR “Fusobacterium nucleatum” OR “Prevotella intermedia”) AND (“Alzheimer” OR “dementia”). This strategy was adapted to fit the search rules of each database and the algorithms can be viewed in Supplementary Table 1. For pertinent articles, their reference lists and all publications that have cited them were also screened for a more comprehensive search. Unpublished work and grey literature were not searched.

Study selection and eligibility criteria

All publications identified by the search were exported into Endnote. Following removal of duplicates, two reviewers (S. L. and R. Z.) screened titles and abstracts of all included publications before full-text reading. A third reviewer (S.D.) assessed any borderline cases for which either inclusion or exclusion could be argued. Studies were included if they were: 1) presented as original articles published in English; 2) aimed to investigate the association between oral bacteria and AD; 3) contained human participants and/or human samples. Case reports, reviews, expert opinions, letters, short commentaries, conference abstracts, and articles only reporting animal, in vitro, or in silico studies were excluded.

Data extraction

The data collection form was piloted with a subgroup of included articles before finalization. For each included article, the following information was extracted independently by two reviewers (S.L. and R.Z.): first author name, year of publication, country, study design, population characteristics, grouping, diagnostic criteria for AD, inclusion and/or exclusion criteria, matched or adjusted confounders, sampling type, targeted bacteria, method for detection, reported outcome type, and major findings. Authors were contacted where study data required to beclarified.

Methodological quality assessment

The quality of included studies was assessed by two reviewers (S.L. and R.Z.) independently using a scoring system adapted from that of Lievense et al. [17] (Table 1). This system contains items for both internal validity and informativeness. For each item, three response options were provided: positive (1), negative (0), and unclear (?). For each study, the possible maximum score (100%) depended on all the criteria applicable to its study design. In case the two reviewers disagreed and failed to reach consensus, a third reviewer (S. D.) was available for a final judgement.

Table 1

Table1

First author, year	Selection before disease was present or at uniform point (CH/CC/CS)	Cases and controls were drawn from the same population (CC)	Partici-pation rate ≥80% for cases/ cohort (CH/CC/CS)	Participation rate ≥80% for controls (CC)	*Sufficient description of subject characteristics (CH/CC/CS)	Exposure was blinded (CH/CC/CS)	Exposure was measured identical for cases and controls (CC)	Exposure was assessed prior to the outcome (CH/CC/CS)	AD was assessed identical in studied population (CH/CC/CS)	AD was assessed reproducibly (CH/CC/CS)	AD was assessed according to standard definitions (CH/CC/CS)	Prospective design was used (CH/CC)	Follow-up time ≥3 years (CH)	Withdrawals ≤20% (CH)	*Infor-mation about completers versus withdrawals (CH)	*Frequ-encies about most important outcome were given (CH/CC/CS)	Appro-priate analysis techniques were used (CH/CC/CS)	Adjusted for at least age and sex (CH/CC/CS)	Individual score	Total obtainable	Total score (%)
Cohort study
Ide 2016 [8]	1	1	1	NA	1	1	NA	1	1	1	1	1	0	1	1	0	1	1	12	13	92
Case-cohort study
Noble 2014 [20]	1	1	1	1	1	1	1	1	1	1	1	0	NA	NA	NA	1	1	1	12	13	92
Nested case-control study
Sparks 2012 [19]	1	1	1	1	1	1	1	1	1	1	1	0	NA	NA	NA	0	1	1	12	13	92
Beydoun 2020 [22]	0	1	1	1	1	1	1	1	0	1	1	0	NA	NA	NA	1	1	1	10	13	77
Case-control study
Panzarella 2020 [27]	0	1	1	1	1	1	1	0	1	1	1	0	NA	NA	NA	0	1	1	10	13	77
Kamer 2009 [18]	0	?	1	1	1	1	1	0	1	1	1	0	NA	NA	NA	1	1	1	9	13	69
Dominy 2019 [21]	0	?	1	1	1	1	1	0	1	1	1	0	NA	NA	NA	1	1	1	9	13	69
Liu 2019 [29]	0	1	1	1	1	1	1	0	0	1	1	0	NA	NA	NA	NA	1	1	9	13	69
Bathini 2020 [25]	0	1	1	1	1	1	1	0	1	1	0	0	NA	NA	NA	NA	1	1	9	13	69
Wu 2021 [30]	0	1	1	1	0	1	1	0	0	1	1	0	NA	NA	NA	NA	1	1	9	13	69
Holmer 2021 [28]	0	1	1	1	1	1	1	0	0	1	1	0	NA	NA	NA	NA	1	1	9	13	69
Riviere 2002 [17]	0	?	1	1	0	1	1	0	1	1	1	0	NA	NA	NA	1	1	0	8	13	62
Poole 2013 [24]	0	?	1	1	0	1	1	0	1	1	1	0	NA	NA	NA	1	1	0	8	13	62
Cirstea [23]	0	1	1	1	1	1	1	0	0	1	1	0	NA	NA	NA	NA	1	1	9	13	69
Guo 2021 [31]	0	1	1	1	1	1	1	0	0	1	1	0	NA	NA	NA	NA	1	1	9	13	69
Cross-sectional study
Leblhuber 2020 [26]	1	?	1	NA	1	1	NA	0	1	1	1	0	NA	NA	NA	1	1	0	7	11	64

CH, cohort; CC, case-control; CS, cross-sectional; NA, not applicable. * informativity item. Not included in the analysis.

Best evidence synthesis

Due to high heterogeneity among the included studies, a meta-analysis was not done in most cases. In this context, we performed a “best evidence synthesis” to summarize the results. According to Lievense et al. [17], the “best evidence synthesis” method was based on the combination of the quality score for internal validity of a study which represents its methodological quality with its study design type. A prospective cohort study was considered the preferred design, followed by a case– control study, and then by a cross-sectional study. The criteria of best evidence synthesis were shown in Table 2. High quality was defined as scoring above the mean value of all scores.

Table 2

Best evidence synthesis based on the included studies

Evidence level	Definition	Association between AD and oral bacteria	Oral bacteria as a risk factor for AD
Strong evidence	Generally consistent findings in multiple high quality cohort studies
Moderate evidence	Generally consistent findings in one high quality cohort study and two or more high quality case-control studies; or in three or more high quality case-control studies	Sparks 2012 [21],Noble 2014 [22],Beydoun 2020 [24],Panzarella 2020 [29]	Sparks 2012 [21],Noble 2014 [22],Beydoun 2020 [24]
Limited evidence	Generally consistent findings in a single cohort study; or in ≤2 case-control studies
Conflicting evidence	Conflicting findings (<75% of the studies reported consistent findings)
No evidence	No studies could be found.

Fig. 1

Flow diagram of literature search and study selection

Statistical analysis

When necessary data were available, meta-analysis was conducted in Review Manager 5.3 software. Differences of oral bacteria between AD patients and control subjects were expressed as weighted mean differences (WMD) and 95% confidence interval (CI) for continuous outcomes, and odds ratio (OR) and 95% CI for dichotomous outcomes. A random effect model was used in all analyses. The Q statistic and I² statistic were used to examine heterogeneity between included studies. Accordingly, a p value <0.1 indicated the presence of heterogeneity in Q statistic. The interpretation of I² values was achieved with the following thresholds: 0–30% (low heterogeneity), 30–60% (moderate heterogeneity), >60% (substantial heterogeneity). Cohen’s kappa coefficient was calculated to assess inter-reviewer reliability (S.L. and R.Z.) for literature screening and methodological quality of the included studies. Kappa values between 0.81 and 1.00 were interpreted as high agreement [18].

RESULTS

Search and selection results

The search process was illustrated in Fig. 1. A total of 1,204 records were identified and narrowed to 638 after removing duplicates. By screening titles and abstracts, 32 publications were selected for full-text reading. Cohen’s kappa for inter-reviewer agreement of initial screening and full text reading were 0.882 and 0.879, respectively, indicating a high level of agreement between the reviewers. Of these, 17 were excluded for the following reasons: 1) 8 studies did not specifically address oral bacteria; 2) 5 studies did not contain an AD group; 3) 2 studies did not provide enough data; 4) 1 study recruited patients with other forms of dementia as control; 5) the full text of 1 study was not available. One study was retrieved by additional hand searching of the bibliographical lists of the selected articles, which yielded a total of 16 articles included in the systematic review. Table 3 presents the characteristics and major findings of the included studies.

Table 3

Characteristics and findings of the included studies

First author, year	Country	Study design (study type; source population)	Demog-raphics (age; sex; race)	AD group	Diagno-stic criteria for AD	Other groups	Matched or adjusted confounders	Inclusion/Exclusion criteria	Sampling type	Detection technique	Targeted oral bacteria	Reported outcome	Major findings
Studies focusing on peridontal bacteria
Riviere 2002 [19]	USA	Case-control; not applicable	55–88 y; 47.1% female; not mentioned	16	Not mentioned	Control: 18	No	Not mentioned	Cortex	Species-specific PCR, IHC	Treponema. amylovorum, T. maltophilum, T. medium, T. vincentii, T. denticola, T. pectinovorum, T. socranskii, Actinobacillus actinomycetemcomitans, Eikenella corrodens, Porphyromonas gingivalis, Prevotella intermedia	Detection rate	• PCR results showed that 14/16 AD cases had at least one Treponema species compared to 4/18 controls (p < 0.001), and AD samples had more Treponema species than controls (p < 0.001).• IHC results showed that 15/16 AD cases had at least one spirochete compared to 7/18 controls (p < 0.01).• Species -specific antigens of Treponema pectinovorum and/or Treponema socranskii were detected in 15/16 AD cases and 6/18 controls (p < 0.01).
Poole 2013 [26]	UK	Case-control; not applicable	77.4±15.7 y; not mentioned; not mentioned	10	NIA-AA	Control: 10	Age	Not mentioned	Cortex	IF, immunoblotting	P. gingivalis, T. denticola, Tannerella forsythia	Detection rate	IF revealed strong cellular surface membrane labeling in 4/10 AD cases, and immunoblotting demonstrated bands corresponding to LPS from P. gingivalis in the same four AD brains. All controls remained negative throughout while the same four cases were consistently positive for P. gingivalis LPS (p = 0.029).
Dominy 2019 [23]	USA	Case-control; not applicable	72.6±13.7 y; 39.7% female; not mentioned	58 brain donors; 10 clinical patients	NIA-AA	Control: 58 brain donors	Age, sex	Not mentioned	Cortex, hippocampus, CSF, saliva	IHC,WB,qPCR	P. gingivalis	Detection rate, bacterial load	• The loads of two P. gingivalis gingipains, RgpB and Kgp (positive in 51/53 and 49/54 AD samples, respectively) were higher in AD brains than control brains (p < 0.001). •P. gingivalis specific gene hmuY was detected in 3/3 AD brains and 5/6 control brains, and in the CSF of 7/10 and saliva of 10/10 clinically diagnosed AD patients.
Kamer 2009 [20]	USA	Case-control; hospital-based	≥40 y; 85.5% female; 82.4% Caucasian	18	NINCDS-ADRDA; DSM IV	Control: 16	Age, sex, APOE genotype, race, education	Exclusion: individuals with other cognition affecting disesases, with psychiatric or behavioral conditions (e.g., depression or substance abuse), or taking medications that affect cognition (e.g., benzodiazepines)	Plasma	ELISA	P. gingivalis, T. forsythia, A. actinomycetemcomitans	Detection rate, number of positive IgG tests	• AD subjects had higher number of positive IgG antibody tests (1.11±0.9) against periodontal bacteria (A. actinomycetemcomitans, P. gingivalis or/and T. forsythia) compared to controls (0.5±0.73) (p = 0.04). • Plasma samples from 72% of AD subjects were positive for at least one antibody to the three bacteria tested compared to 38% of controls (p = 0.04)
Sparks 2012 [21]	USA	Nested case-control; population-based	71.5±6.8 y; 59.5% female; Caucasian 99.4%	35	NINCDS-ADRDA	MCI: 46; control: 77	Age, sex, race, educaton, occupation, smoking, baseline MMSE score, follow-up, diabetes	Inclusion: BRAINS participants who were cognitively intact at their baseline serum draw and had available serial blood draws	Serum	ELISA	P. gingivalis, T. forsythia, A. actinomycetemcomitans, T. denticola, Campylobacter rectus, P. intermedia, Fusobacterium nuleatum	IgG antibody levels presented by diagram	Antibody levels against F. nucleatum and P. intermedia were increased (p < 0.05) at baseline serum draw in AD subjects compared to controls.
Noble 2014 [22]	USA	Case-cohort; community-based	75.5±6.8 y; 67.6% female; 48% Hispanics, 27% non-Hispanic blacks, 26% non-Hispanic whites	110	NINCDS-ADRDA	Control: 109	Age, sex, APOE genotype, race, education, smoking, diabetes, hypertension, hyperlipidemia, heart disease, stroke histroy	Inclusion: WHICAP participants enrolled in 1999–2000 with no prevalent cognitive impairment or dementia at the time of phlebotomy and ≥1 follow-up visit thereafter. Control: absence of clinical dementia at the time of the last available follow-up	Serum	Checker-board immunob-lotting	P. gingivalis, T. forsythia, A. actinomycetemcomitans, T. denticola, C. rectus, Eubacterium nodatum, Actinomyces naeslundii	Distri-bution of IgG antibody levels	• High anti-A. naeslundii titer (>640 ng/ml, present in 10% of subjects) was associated with increased risk of AD (HR = 52.0, 95% CI: 1.1–3.8). • High anti-E. nodatum IgG (>1755 ng/ml;19% of subjects) was associated with lower risk of AD (HR = 50.5, 95% CI: 0.2–0.9).
Ide 2016 [8]	UK	Prospective cohort; community-based	77.7±8.6 y; 50.0% female; not mentioned	60	NINCDS-ADRDA	No	No	Inclusion: non-smoking patients with mild to moderate dementia and a minimum of 10 teeth who had not received treatment for periodontitis in the past 6 months; those who had been on a stable cholinesterase inhibitor for at least 6 months; modified Hachinski Ischaemic Scale score >4	Serum	ELISA	P. gingivalis	IgG antibody levels	There was no significant association between baseline serum P. gingivalis antibody levels and the rate of decline on the ADAS score (p = 0.9) or rate of decline on the MMSE score (p = 0.1).
Beydoun 2020 [24]	USA	Nested case-control; population-based	63.7±10.4 y; 55.9 % female; 82.9% non-Hispanic white, 8.3% non-Hispanic black, 2.8% Mexican-American and 6.1% other	990	ICD-9	Incident all-cause dementia: 1,737; control: 3,923	Age, sex, race, education, poverty income ratio, place of residence, household size, marital status, nutrition, smoking, drug use, alcohol, physical activity, self-rated health, comorbidity index, allostatic load, weight status, dental status, social support variables	Not specified	Serum	Checker-board immuno-blotting	T. denticola, T. forsythia, P. gingivalis, P. intermedia, A. actinomycetemcomitans, F. nucleatum, C. retus, E. nodatum, P. nigrescens, P. melaninogenica, Parvimons micra aka Micromonas micros, Selenomonas noxia, E. corrodens, Capnocylophaga ochracea, Streptococcus intermedius, S. oralis, S. mutans, Vellonella Parvula, A. naeslundii	IgG antibody levels	• Increased AD mortality was associated with higher IgG titers against A. naeslundii, P. gingivalis, P. intermedia, P. melaninogenica, P. nigrescens, S. intermedius, S. oralis and T. forsythia (p < 0.033). • Increased AD incidence was associated with a composite of higher anti-C. rectus and anti-P. gingivalis IgG titers (p = 0.012). • Increased AD mortality was increased with a composite (p = 0.017) loading highly on IgG against P.gingivalis, P. intermedia, P. nigrescens, F. nucleatum, C. rectus, S. intermedius, C. Ochracea, and P. melaninogenica.
Leblhuber 2020 [28]	Austria	Cross-sectional; hospital-based	78.1±2.2 y; 45% female; not mentioned	20	ICD-10	No	No	Not mentioned	Alveolar fluid	RT-qPCR	A. actinomycetemcomitans, T. denticola, T. forsythia, P. gingivalis, P. intermedia	Detection rate	• Detection rate in AD subjects: T. forsythia 35% (7/20), T. denticola 30% (6/20), P. gingivalis 25% (5/20), P. intermedia 0% (0/20), A. actinomycetemcomitans 0% (0/20) • AD subjects positive for P. gingivalis showed statistically lower MMSE scores (p < 0.05).
Panzarella 2020 [29]	Italy	Case-control; community-based	80±8.68 y, 55.0% female, not mentioned	20	NIA-AA	MCI: 20; control: 20	Age, sex, APOE genotype, education, BMI, smoking, alcohol, hypertension, diabetes, dyslipidemia, atrial fibrillation, high-sensitivity C-reactive protein	Not specified	Sub-gingival plaque	RT-qPCR	A. actinomycetemcomitans, F. nucleatum, T. denticola, T. forsythia, P. gingivalis, P. intermedia	Bacterial load	A higher F. nucleatum load was detected in AD than in control (p = 0.02).
Liu 2019 [31]	China	Case-control, hospital-based	64.2±9.6 y; 50.0% female; not mentioned	39	NINCDS-ADRDA	Control: 39	Age, sex	Exlusion: other causes of dementia or other types of dementia; any kind of other neurodegenerative diseases (e.g., Parkinson’s disease); severe cardiac, pulmonary, hepatic, renal diseases, any kinds of tumor, rheumatoid immune-related diseases (e.g., Sjogren’s syndrome); a history of smoking, alcohol drinking, taking antibiotics, glucocorticoids, or probiotics within one month; oral surgery or dental procedure, inflammation of oral or perioral tissues, and other chronic disease in the oral cavity within 2 months.	Saliva	16 S rRNA sequencing Database: RDPRegion: V3-V4Clustering method: OTU	Salivary microbiota	Bacterial diversity and composition	Relative abundance of Moraxella, Leptotrichia and Sphaerochaeta in the saliva of AD was greatly increased, whereas that of Rothia was significantly reduced.
Bathini 2020 [27]	Switzer-land	Case-control; hospital-based	69.7±8.8 y; 57.5% female; 100% Caucasian	17	MMSE <20 and CDR >1	MCI: 21; CNr: 15; CNh: 27	Age, sex, APOE genotype, race, education, BMI	Inclusion: Control: MMSE score ≥26; Exclusion: Control: general anesthesia in the last 3 months; severe sensory or motor deficits; severe systemic diseases; current psychiatric problems. history of central nervous system diseases; head injury with loss of consciousness for more than 5 min; psychiatric disorders requiring hospitalization, intake of potent psychoactive drugs. diabetes. AD: >10 at the Geriatric Depression Scale and findings showing other reasons for cognitive problems; severe systemic diseases; psychiatric disorders requiring hospitalization, intake of potent psychoactive drugs; diabetes.	Saliva	16 S rRNA sequencingDatabase: not mentionedRegion: V3-V4Clustering method: NM	Salivary microbiota	Bacterial diversity and composition	•Filifactor villosus decreased with the increasing severity of AD (p < 0.05). •F. villosus and P. tannerae were the best differentiators between CNh and AD.•Positive interactions among P. gingivalis and Filifactor species were detected in CNh, MCI, and AD.
Wu 2020 [32]	Taiwan	Case-control, not mentioned	71.4±19.9 y; 65.7% female; 100% Asian	17	NIA-AA	Control: 18	Age, sex, DMFT index	Inclusion: middle-class native Taiwanese. Control: individuals who visited dental clinics for dental health check-ups and had no cognitive complaints. Exclusion: confounding neurological (e.g., Parkinson syndrome) diseases, oral cancer diseases (e.g., oral squamous cell carcinoma, osteosarcoma, and salivary gland cancer), antibiotic administration within the previous two months.	Supra-gingival plaque	16 S rDNA sequencingDatabase: HOMD, 16 S RefSeq Version 15.2Region: V1–V9Clustering method: OTU	Supra-gingival microbiota	Bacterial diversity and composition	•Fusobacterium, Cardiobacterium and Porphyromonas significantly decreased while Lactobacillus, Streptococcus, Actinomyces and Veillonella were enriched in AD subjects versus controls.
Holmer 2021 [30]	Sweden	Case-control; population-based	50–80 y; 52.4% female; not mentioned	46	NIA-AA	MCI: 40; SCI: 46; control: 63	Age, sex, place of birth, education, income, marital status, smoking, diabetes, number of comorbidities, number of medications, tooth number, marginal alveolar bone loss	Inclusion: Control: individuals who had not sought medical attention for memory loss or experienced memory loss; MMSE score ≥28 and passing the CDR test. Exclusion: severe medical conditions (brain tumours, clinically significant liver, kidney or lung dysfunction, chronic inflammatory disease and psychiatric diseases	Sub-gingival plaque	16 S rRNA sequencingDatabase: HOMDRegion: V3-V4Clustering method: OTU	Sub-gingival microbiota	Bacterial diversity and composition	Two OTUs were particularly abundant in AD subjects compared to controls: Slackia exigua, which was also associated with deep periodontal pockets, and a Lachnospiraceae bacterium.
Cirstea 2022 [25]	Canada	Case-control; hospital-based	65–78 y; 33.3% female; not mentioned	45	NIA-AA	Control: 54	Age,sex, smoking, serum CRP, ever lived rural	Patient inclusion: 1) age 50–85 y, 2) a diagnosis of probable AD dementia, or dementia of mixed AD/vascular etiology, based on the McKhann criteria [14], 3) age of onset between 50–80 y. Exclusion: 1) a diagnosis of a non-AD dementia, 2) a concurrent diagnosis of a gastroinflammatory disease, 3) antibiotic use within the previous 3 months, 4) probiotic use within the previous month, 5) a current illness, including a cold, flu, or bacterial infection.Control Inclusion: age 50–85.Exclusion: the same as for AD patients, with the additions of 1) a current diagnosis of AD or a related neurological disorder, 2) a known family history of early-onset AD or related disorder, 3) a score lower than 26 on the Montreal Cognitive Assessment.	Oral swabs	16 S rDNA sequencing Database: Greegenes, BLASTRegion: V4Clustering method: ASV	Oral microbiota on the buccal and sublingual mucosa	Bacterial diversity and composition	* AD subjects had a lower abundance of the phylum Firmicutes (driven largely by the family Streptococcaceae), the family Actinomycetaceae and the species Rothia mucilaginosa, and a higher abundance of the family Weeksellaceae versus controls.* The relative abundance of P. gingivalis was far more prevalent in AD, found in 13/45 patients (28.9%) versus 3/54 controls (5.6%) (p = 0.0022).
Guo 2021 [33]	China	Case-control; hospital-based	71±7.28 y, 63.5% female, not mentioned	26	ICD-10	26	Age, sex, BMI, diabetes, hypertension, hyperlipidemia, tooth number, residual root, periodontal status indexes, periodontitis severity	Inclusion (a) AD diagnosed according to ICD-10; a cognitively intact status diagnosed with the value of the MMSE greater than 27; (b) number of teeth greater than 7; (c) able to cooperate during oral examination; (d) subjects or their legal guardians signing the informed consent forms.Exclusion: (a) diagnosed with other types of dementia, such as vascular dementia, Lewy body dementia and alcoholic dementia; (b) value of the HAMD more than 6; (c) teeth number less than 7; (d) receiving supragingival scaling, subgingival scaling and root planning and other periodontal surgery treatments during the 6 months before sampling; (e) havingopen surgical treatments of head and/or mouth; (f) diagnosed with acquired immune deficiency syndrome, hepatitis B infection, hepatitisC infection and other infectious diseases; (g) a history of cancer, staphylococcal toxic shock syndrome, inherited metabolic diseases, autoimmune diseases, hormone-dependent diseases, radiotherapyand chemotherapy; (h) taking antibiotics, immunomodulators, cytokines and probiotics within the last 3 months.	SalivaGCF	16 S rRNA sequencingDatabase: NCBIRegion V1–V9Clustering method: ASV	Salivary microbiotaPeriodontal microbiota	Bacterial diversity and composition	The predominant bacterium of salivary microbiome and periodontal microbiomefrom AD patients was Streptococcus oralis* and P. gingivalis, respectively.* The correlation analysis revealed that Veillonella parvula was positively associated with AD in both saliva and GCF and that P. gingivalis was positively associated with AD only in GCF.

PCR, polymerase chain reaction; IHC, immunohistochemistry; NIA-AA, National Institute on Aging and the Alzheimer’s Association; IF, immunofluorescence; LPS, lipopolysaccharide; CSF, cerebrospinal fluid; WB, western blotting; qPCR, quantitative PCR; NINCDS-ADRDA, National Institute of Neurologic and Communicative Disorders and Stroke and Alzheimer’s Disease and Related Disorders Association; DSM IV, Diagnostic and Statistical Manual of Mental Disorders, 4th. Edition; APOE, apolipoprotein E; ELISA, enzyme-linked immunosorbent assay; IgG, immunoglobulin G; MCI, mild cognitive impairment; MMSE, Mini-Mental State Examination; BRAINS, Biologically Resilient Adults in Neurological Studies research program; WHICAP, Washington Heights-Inwood Columbia Aging Project; ADAS, Alzheimer’s Disease Assessment Scale; ICD, International Classification of Diseases; RT-qPCR, quantitative reverse transcription polymerase chain reaction; BMI, body mass index; RDP, Ribosomal Database Project; OTU, Operational Taxonomic Unit; CDR, Clinical Dementia Rating test; CNr, cognitively normal at risk; CNh, cognitively normal healthy; DMFT, the decayed, missing, and filled teeth; HOMD, Human Oral Microbiome Database; SCI, subjective cognitive impairment; ASV, amplicon sequence variant; GCF, gingival crevicular fluid; Note: Findings on bacterial diversity of the microbiome-wide association studies are presented in Table 4.

Description of study characteristics

The 16 studies were published between 2002 and 2022 and were conducted in three regions of the world: 7 in North America [19 –25], 6 in Europe [8 , 26–30], and 3 in East Asia [31 –33]. Of these, 11 had a case-control design [19 , 29–33], 2 were nested case-control [21, 24], 1 was a perspective cohort study [8], 1 was a case-cohort study [22], and 1 was cross-sectional [28]. The approach to participant recruitment was hospital-based in six studies [20 , 33], population-based in three [21 , 30], community-based in three [8 , 29] and not mentioned by one [32], except three postmortem studies using banked brain tissues [19 , 26]. A total of 7,920 adults, mostly between the age of 40–90 years, were enrolled. The percentage of females was 55.7%. Of all participants, AD cases and cognitively normal controls accounted for 19.4% and 56.5% respectively; the left 24.1% were subjects with other forms of dementia or cognitive impairment. Most studies had adjusted age and sex of the participants. Other potential confounders controlled by at least three studies included race, APOE genotype, education, socioeconomic aspects, smoking, body mass index or weight status, dental status, and associated comorbidities like diabetes, cardiac, and vascular diseases.

The diagnosis of AD

The diagnosis criteria for AD varied widely among studies. Five studies [8 , 31] used the well-accepted criteria of National Institute of Neurologic and Communicative Disorders and Stroke and Alzheimer’s Disease and Related Disorders Association published in 1984, of which one [20] also applied the criteria from Diagnostic and Statistical Manual of Mental Disorders [20]. Four studies [25 , 32] used the 2011 clinical criteria and two [23, 26] used the 2012 neuropathological criteria proposed by the National Institute on Aging and the Alzheimer’s Association. Three studies [24 , 33] used the criteria of International Classification of Diseases. In contrast, one [27] classified AD by the scores of mini-mental state examination and clinical dementia rating test. One postmortem study [19] did not report the method of confirming AD in the donated brains and we failed to get this information from either the authors or the supplying brain bank.

Quality assessment and best evidence synthesis

Table 1 presents all the studies stratified by their study design types and in order of their methodological quality scores. The scores ranged from 62% to 92%, with a mean value of 73%. Inter-reviewer agreement for quality assessment was high with a kappa of 0.885 and a weighted kappa of 0.867. Five of the sixteen studies were given a score above the mean and rated as high-quality. These included one prospective cohort study, one case-cohort study, two nested case-control study, and one case-control study. Therefore, the best evidence synthesis suggests a moderate level of evidence for an overall association between oral bacteria and AD. Of the five high-quality studies, four containing a longitudinal component could actually examine the role of oral bacteria as a risk factor for AD, and the findings of three, except the prospective cohort study, were in favor of this role. This means that there is moderate evidence supporting the involvement of oral bacteria in AD development.

Analysis of periodontal bacteria and AD

Of the 16 studies, 10 studies investigated periodontal pathobionts. P. gingivalis was the bacterium most frequently examined (by nine studies), followed by T. denticola, Tannerella forsythia, and A. actinomycetemcomitans (all by seven studies). Studied samples included brain tissues, serum/plasma, and oral samples.

Brain samples

Only three studies investigated the presence of periodontal pathobionts and/or their components in postmortem brains, mainly in cortex samples. Overall oral bacterial infection rate of brain samples was 83.75% in AD patients and 38.46% in controls. As Fig. 2 shows, we found over a ten-fold increased risk of AD when there was a detectable level of periodontal bacteria in the brain (OR = 10.68 95% CI: 4.48–25.43; p < 0.00001) with a pooled sample size of 158 subjects (80 AD patients versus 78 controls). The inter-study heterogeneity was low, given an I² of 0%. Cerebral infection rate of P. gingivalis was 66.25% in AD patients and 33.33% in controls. The risk of AD was over six-fold when P. gingivalis was positive in the brain (OR = 6.84 95% CI: 2.70–17.31; p < 0.0001) (p value for heterogeneity: 0.42, I² = 0%).

Fig. 2

Pooled risk estimates of oral bacteria in the brain and AD risk. Zero values in cells were replaced with one to facilitate calculations of ORs and CIs.

In the study by Riviere et al., six oral Treponeme species, T. denticola, T. pectinovorum, T. amylovorum, T. maltophilum, T. medium, and T. socranskii, were detected by polymerase chain reaction (PCR) in the frontal lobe cortex with at least one species positive in 14 of 16 AD cases compared to 4 of 18 controls (p < 0.001) [19]. AD samples exhibited more Treponema species than controls (p < 0.001). Poole and colleagues observed immunostaining of P. gingivalis in 4 of 10 AD brains and none of 10 controls [26]. Dominy and colleagues reported significantly higher loads of a group of key virulence-associated proteins of P. gingivalis, gingipains, in AD samples than in non-demented controls (p < 0.001), with intraneuronal staining of gingipains in over 90% AD cortex samples as well as in the AD hippocampus [23]. The hmuY gene, which is highly specific for P. gingivalis, was detected in 3 of 3 AD brains and 5 of 6 control brains, and in the CSF of 7 out of 10 clinically diagnosed AD patients.

Serum/plasma samples

Five studies analyzed host immune responses against oral bacteria by analyzing immunoglobulin G (IgG) serum or plasma antibody levels. Either enzyme-linked immunosorbent assay or checkerboard immunoblotting was performed, and the number of targeted bacterial species ranged from 1 to 19 across studies. Higher antibody levels against nine individual species were linked with increased risk of AD, including Actinomyces naeslundii, Fusobacterium nucleatum, P. gingivalis, Prevotella intermedia, P. melaninogenica, P. nigrescens, S. intermedius, Streptococcus oralis, and T. forsythia. Combinations of different species were investigated as well. In the study by Kamer et al., the combination of IgG antibodies against P. gingivalis, T. forsythia, and A. actinomycetemcomitans were positive in 72% AD patients compared to 38% controls (p = 0.04) [20]. Beydoun and colleagues identified four combinations involving antibodies against 11 oral bacteria to be associated with increased AD incidence and/or mortality [24]. By contrast, antibodies against Eubacterium nodatum and A. actinomycetemcomitans were reported negatively related to AD risk [22, 24]. The six-month prospective study by Ide and colleagues reported no association between baseline anti-P. gingivalis IgG levels and the rate of cognitive decline in a cohort of AD patients [8].

Oral samples

Two studies used quantitative reverse transcription PCR to quantify specific bacterial pathobionts in periodontal niches: the study by Leblhuber et al. found an association between the presence of P. gingivalis in alveolar fluid and lower cognitive scores in AD patients (p < 0.05) [28] and the study by Panzerella et al. revealed a higher F. nucleatum load in the sub-gingival plaque of AD patients than in controls (p = 0.02) [29].

Table 4

Summary of microbial diversity assessments from included studies

(a)
Study	α Diversity	Results	p
Liu 2019 [31]	Observed species	a significantly lower diversity in AD patients than in controls	p < 0.001
Chao 1 index	a significantly lower diversity in AD patients than in controls	p < 0.001
Shannon index	a significantly lower diversity in AD patients than in controls	p < 0.001
Phylogenetic Diversity whole tree index	a significantly lower diversity in AD patients than in controls	p < 0.001
Bathini 2020 [27]	Shannon index	no significant difference in diversity between groups	NM
Wu 2020 [32]	Observed species	a lower diversity in AD patients (5569±1115) than in controls (6696±1666)	p < 0.05
Chao 1 index	a lower diversity in AD patients (227.94±61.47) than in controls (236.04±42.65)	p < 0.05
Shannon index	a lower diversity in AD patients (5.38±0.97) than in controls (5.64±0.51)	p≥0.05
Simpson reciprocal index	a lower diversity in AD patients (26.59±12.78) than in controls (29.47±11.56)	p≥0.05
Holmer 2021 [30]	Observed richness	a higher diversity in cases (AD, MCI and SCI combined) than in controls	p = 0.008
a significant difference in richness across AD, MCI, SCI and control groups	p = 0.018
Shannon index	a higher diversity in cases (AD, MCI and SCI combined) than in controls	p = 0.010
a significant difference in diversity across AD, MCI, SCI and control groups	p = 0.002
Cirstea 2022 [25]	Chao 1 index	a higher diversity in AD patients than in controls	p = 0.00073
(b)	p
Study	β Diversity	Results	P values
Liu 2019 [31]	Weighted UniFrac PCoA	no significant difference in the composition of oral microbiota between AD patients and controls	NM
Holmer 2021 [30]	NMDS based on Bray-Curtis dissimilarity	a significant difference between cases (AD, MCI and SCI combined) and controls	p = 0.014
a significant difference across AD, MCI, SCI and control groups	p = 0.041
Cirstea 2022 [25]	Unweighted UniFrac PCoA	a significant difference in the composition of oral microbiota between AD patients and controls	p = 0.001
Jaccard similarity index	a significant difference in the composition of oral microbiota between AD patients and controls	p = 0.001
Bray-Curtis dissimilarity	no significant difference in the composition of oral microbiota between AD patients and controls	p = 0.13
Guo 2021 GCF [33]	Weighted UniFrac- PCoA	a distinction in the composition of oral microbiota between AD patients and controls	NM
Bray-Curtis dissimilarity	a significant difference in the composition of oral microbiota between AD patients and controls	p = 0.003
Guo 2021 Saliva [33]	Weighted UniFrac- PCoA	no significant difference in the composition of oral microbiota between AD patients and controls	NM
Bray-Curtis dissimilarity	no significant difference in the composition of oral microbiota between AD patients and controls	p = 0.055

AD, Alzheimer’s disease; NM, not mentioned; PCoA, Principal Coordinate Analysis; NMDS, Non-metric multidimensional scaling; MCI, mild cognitive impairment; SCI, subjective cognitive impairment.

Analysis of oral bacterial communities and AD

A total of six studies analyzed the diversity and composition of oral bacterial communities in AD patients using genome sequencing methods (Table 3). Studied samples included saliva, supra- and sub-gingival plaque, oral mucosa swabs, and gingival crevicular fluid.

Alpha diversity and beta diversity

Five studies reported the alpha diversity and four reported the beta diversity of studied bacterial communities. As summarized in Table 4, findings were inconsistent across studies. Compared to healthy controls, the alpha diversity of oral microbiota in AD patients was lower in two studies but was significantly higher in one study and higher in AD, mild cognitive impairment, and subjective cognitive impairment combined cases in one study. Five alpha diversity indices were assessed in total, where Shannon index was most frequently reported. Due to high heterogeneity among studies, statistical analysis was conducted only with Shannon index [WMD=–0.52, 95% CI (–0.72, –0.31), p < 0.00001] (p value for heterogeneity: 0.49, I² = 0%) (Fig. 3), indicating a lower diversity of oral microbiota in AD patients than in controls. Regarding beta diversity, the overall composition of salivary microbiota did not differ between AD patients and controls based on two studies, while three studies revealed a significant difference in the microbial composition of other niches in the oral cavity between cases andcontrols.

Fig. 3

Difference in Shannon index of oral microbiota between AD patients and controls.

Relative abundance of bacterial taxa

Data on bacterial abundance were analyzed in terms of different taxonomic levels (Fig. 4). In the phylum level, AD patients and healthy controls showed similar abundance of Firmicutes (p = 0.13). In the family level, similar abundance of Streptococcaceae (p = 0.97) and Actinomycetaceae (p = 0.97) was observed between AD patients and controls. In the genus level, a significantly decreased abundance of Rothia was observed in AD patients versus controls with an WMD of –1.93% [95% CI (–2.44%, –1.42%), p < 0.00001] (p value for heterogeneity:0.75, I² =0%). No significant difference was found in the abundance of Fusobacterium (p = 0.61), Streptococcus (p = 0.47), Porphyromonas (p = 0.59), Actinomyces (p = 0.28), Prevotella (p = 0.18), and Veillonella (p = 0.67) between AD patients and controls. In the species level, the abundance of Rothia Mucilaginosa was not obviously different (p = 0.73) between AD patients and controls.

Fig. 4

Differences in relative abundance of oral bacterial taxa between AD patients and controls (%).

Comparison by region

There is evidence that the composition and diversity of human microbiome differ with geographic location [34]. Considering this, we made a comparison between findings from East Asia, Europe and North America where applicable (Fig. 5 and Supplementary Table 2). With regard to oral microbial diversity, Shannon index remained to be decreased in AD patients versus healthy controls in both East Asia [WMD=–0.51, 95% CI (–0.80, –0.21), p = 0.0008] and Europe [WMD=–0.45, 95% CI (–0.90, –0.00), p = 0.05]. However, changes of bacterial relative abundance were inconsistent across regions. Compared to healthy controls, AD patients from East Asia exhibited more abundance of Firmicutes, Streptococcaceae [WMD = 2.95, 95% CI (1.11, 4.79), p = 0.002] and Streptococcus [WMD = 2.07, 95% CI (0.98, 3.15), p = 0.0002], but such trend was not found in North American AD patients for Firmicutes (p = 0.61) and Streptococcaceae (p = 0.24), and a contrary pattern was observed in European AD patients for Streptococcaceae [WMD=–2.21, 95% CI (–4.36, –0.06), p = 0.04]. While the abundance of Actinomycetaceae, Rothia, Actinomyces, and Rothia Mucilaginosa was not significantly different between AD patients and controls in the overall comparison, all of them were significantly decreased in one of the three continents.

Fig. 5

Changes in relative abundance of bacterial taxa in AD patients as compared to controls categorized by region of the included microbiome-wide association studies.

DISCUSSION

This systematic review and meta-analysis comprehensively investigated the association between oral bacteria, including both individual species and bacterial communities, and AD. Qualitative synthesis suggested a moderate level of evidence for oral bacteria having a role in AD development. In terms of individual species, ten studies examined periodontal bacteria in AD patients (and controls) at three different levels: the brain, serum, and oral cavity. We found over a ten-fold and six-fold increased occurrence of AD when periodontal pathobionts and P. gingivalis were detected in the brain, respectively. While the diversity of oral microbiota tended to decline in AD patients versus healthy controls, extant evidence does not support a consistent distinct composition of oral microbiota in AD.

Ten of the sixteen included studies had targeted periodontal bacteria and nine of them investigated P. gingivalis in AD. P. gingivalis, a Gram-negative anaerobic bacillus colonizing subgingival biofilms, is considered the key pathobiont for periodontitis and has been increasingly investigated as the possible mechanistic link between periodontitis and AD over the past decade. This pathobiont has a large arsenal of virulence factors whereby it manipulates host immune responses, pushing it towards a nonspecific, sustained inflammatory response that benefits the bacterium and damages not only the periodontal apparatus but remote organs after systemic dissemination through injured periodontal epithelia [35, 36]. The postmortem studies in our review [23, 26] reported a higher abundance of P. gingivalis in the brain samples of AD patients compared to non-demented controls, and we found that its intracerebral presence was related to a six-fold increased risk of AD. In vivo studies have demonstrated that P. gingivalis infiltrates murine brains after oral infection and induces various AD-type changes like neuroinflammation, Aβ accumulation, tau tangles, neurodegeneration, and cognitive impairment [37 –42]. Together, these findings indicate a pathogenic role for P. gingivalis in the AD brain. Besides, in vitro experiments have confirmed that P. gingivalis invades and persists within neurons which display signs of AD-like pathology [43]. Three components, LPS, gingipains, and nucleic acids of P. gingivalis, were detected in the included postmortem studies. Specifically, P. gingivalis LPS was located on cell surface membranes and extracellularly in the brain while gingipains were observed within neurons near tau tangles and Aβ, possibly reflecting distinct contributions of the two molecules to AD neuropathology. As with other Gram-negative bacteria, LPS is a major structural component of the outer membrane of P. gingivalis, which can be recognized by the pattern recognition receptors expressed on host immune cells [36, 44]. P. gingivalis LPS is reported to trigger the release of a number of pro-inflammatory cytokines and chemokines by activating toll-like receptors on microglia [45 –48], the brain innate immune cells, as well as on brain endothelial cells [49], hence implicated in the neuroinflammatory process of AD. Gingipains are cysteine proteases essential for the survival and pathogenicity of P. gingivalis. In vitro gingipains degrade tau proteins in neurons and some peptide products are involved in the formation of NFT [23, 43]. Recently, DNA extracts from P. gingivalis have been demonstrated to trigger the aggregation of both Aβ_1 - 42 and tau protein, which suggests a new role for extracellular DNA of this bacterium in the pathogenesis of AD [51, 52]. Taken together, the observations in postmortem brains are in harmony with previous preclinical evidence to support a contribution of P. gingivalis towards AD development through unique modes of action entailing its LPS, gingipains, and DNA. Further, oral administration of gingipain inhibitors effectively reduced neuroinflammation and the loss of hippocampal neurons in P. gingivalis-infected mouse brains, highlighting P. gingivalis gingipains as a potential target for treating both periodontal diseases andAD [23].

In Riviere and colleagues’ study, six oral spirochetes, including T. denticola, were identified in significantly more AD brains than controls [19]. Like P. gingivalis, T. denticola is a major pathobiont for periodontitis whose relative abundance in subgingival plaque rises sharply before the active progression of periodontitis [52]. T. denticola and P. gingivalis act synergistically in bacterial survival, growth, biofilm formation, and pathogenicity [53 –58]. They predominantly exist in deep periodontal pockets at depths over 4 mm and form microcolony blooms at the outer layer of sub-gingival plaques adjacent to the pocket base, which implicates the mutualistic symbiosis of the two major pathobionts in periodontitis progression [59]. As P. gingivalis and T. denticola were both detected in human brains, the question arises whether their pathogenic synergy in sub-gingival biofilms is involved in the pathogenesis of AD in the brain. Looking at the seven clinical studies targeting periodontal bacteria in our review, five of them reported more than one individual species or combinations of different pathobionts to be associated with altered risk of AD, which expands the topic into polymicrobial involvement and interactions in AD. Periodontitis and AD are both multifactorial chronic diseases that have been linked with periodontal bacteria and inflammation [60]. In an analogous way to periodontitis which results from the mutual positive feedback between dysbiosis of sub-gingival polymicrobial communities and local excess, uncontrolled and chronic inflammation, it has been proposed that AD may be at least in part driven by the interplay between dysbiotic polymicrobial communities and inflammation in the brain [61]. However, the basic premise underlying this argument is that bacterial pathobionts and commensals can reside permanently in the brain, which remains to be evidenced by direct identification of live bacterial cell bodies in the brain rather than bacterial components which can be either residues after host phagocytosis or toxins secreted as free molecules or transported by outer membrane vesicles when entering the brain.

It has been shown that higher anti-P. gingivalis IgG titers are associated with cognitive decline in the cognitively normal population [62]. Three of the five serum-based studies in our review had sampled their participants at baseline before the clinical symptoms of dementia [21 , 24]. All of them found that higher baseline IgG levels against specific oral bacteria were associated with higher AD risk. From a temporal point of view, the manifestation of AD can be divided into three broad stages, i.e., preclinical AD, mild cognitive impairment, and dementia due to AD [1]. The biologic onset of AD occurs prior to its clinical stage for over two decades [63]. This means that if oral bacteria are responsible for the neuropathological changes of AD, they need to access and colonize the brain at a time point over 20 years before the clinical emergence of AD. In this case, the increased baseline serum antibodies of AD patients may reflect the hematogenous spread of oral bacteria which ends up in the brain initiating AD lesions. However, the longest mean follow-up time of the three studies was about 16.7 years, probably not long enough to encompass the whole preclinical phase of AD. In brief, the spatiotemporal nexus between oral bacteria and AD, as a precondition for causality, remains to be deciphered.

Overall, findings were inconsistent across the six oral microbiome-wide association studies. This can be partially explained by the wide variation in sampling sites. The oral ecosystem is highly intricate as it contains multiple unique niches, such as saliva, surfaces of oral mucosa and tongue, supra-gingival and sub-gingival plaques. The microenvironment of each niche varies with the surface structure, local pH, exposure to oxygen, availability of nutrients, among others, thus harboring distinct microbial communities. Among the six included studies, only Guo et al. compared oral microbiota changes in two sites of the oral cavity, viz. saliva and gingival crevicular fluid. They reported that diversity alteration and predominant species varied between salivary and periodontal microbiota [33]. Moreover, periodontal microbiota was more sensitive to cognitive decline than salivary microbiota. Considering that intraoral bacterial profiles are broadly different across different niches in healthy subjects [64], these profiles presumably experience site-specific changes during AD development. Study region is another potential confounder. When dividing the six studies by region, AD patients from East Asia, Europe, and North America showed divergent changing patterns of microbiome structure as compared with controls. This divergency might result from contextual differences of the study populations, such as race, genetics, geographical, social, and cultural features [34], but data must be interpreted with caution. Notably, the interplay between oral microbiota, host response and AD probably occurs throughout the continuum of AD, which can be quite complex. This complexity arises not only from the diverse oral bacterial taxa colonizing distinct niches of the oral cavity and the broad individual differences in the repertoire of these taxa, but also from the intricacy in host immune responses and the multifaceted nature of AD [15]. Future microbiome-wide association studies with more different populations and larger samples are warranted to investigate the site-specific changes of oral bacterial communities over the whole AD spectrum. In addition to compositional studies, future transcriptomics and proteomics analysis may deepen our understanding of the functional interrelationship between AD and the oral ecosystem.

Compared to non-demented controls, our meta-analysis revealed a decline in the alpha diversity of oral microbiota in AD patients. Interestingly, a recent meta-analysis reported a reduced alpha diversity in the gut microbiota of AD patients as compared to controls [65]. Oral and gut bacteria may have close interactions as the oral cavity and gut are interconnected through the alimentary canal. Gut microbial dysbiosis has been shown to promote cognitive disorders and AD pathologies in the brain [66, 67]. As inter-organ microbial network is emerging as an important regulator in human pathophysiological processes [68], we advocate for future work to examine if an oral-gut microbiome axis exists in the pathogenesis of AD.

Our literature research has several limitations. First, a small number of studies were identified, mostly with a case-control design; only one prospective cohort study was included, while its follow-up time was only 6 months. This is insufficient to make causal inferences between oral bacteria and AD. Second, there was high heterogeneity in the methodology and results of the included studies. Sampling type may be a key contributor to the heterogeneity across studies; Although most studies employed PCR and/or immunological methods for bacterial detection, technical details including primer designs, DNA or antibody preparation, threshold setting and time frames might diversify the results. As for high-throughput sequencing, the inconsistency in results may also derive from differences in sample sizes, targeted 16 S rDNA hypervariable region, the adapted sequencing and analyzing strategies; the diagnostic criteria for AD and the diseased stages of patients vary among studies. Clearly, standardization for protocols, techniques, and AD definition are required for more conclusive evidence about the relationship between oral bacteria and AD. The small sample size in most studies further increased the heterogeneity not only among studies as coupled with geographical differences but within the study itself. All these factors had restricted our statistical analysis to very few studies, indicating that our data may be preliminary and underpowered. We only included articles written in English, which may have excluded eligible articles with useful data in other languages. However, to our knowledge, this is the first systematic review and meta-analysis that presents historic and new data on the association between not only individual bacteria but also bacterial communities in the oral cavity and AD.

Conclusion

Our literature review showed a moderate level of evidence for an overall association between oral bacteria and AD. Our meta-analysis demonstrated an increased risk of AD when periodontal pathobionts were detectable in the brain and a decline in oral microbial diversity during AD. These findings lend support to the infectious hypothesis of AD. However, given the inconsistencies across studies, well-designed and well-reported prospective studies in large cohorts are in demand to test the role of oral bacterial pathobionts in the development of AD. Additional studies are warranted to decipher the changing of oral microbial patterns during AD continuum so as to illuminate the preventive, diagnostic and therapeutic potential of oral microbiota in AD.

Footnotes

ACKNOWLEDGMENTS

This study was supported by funds from the Postdoctoral Program of the First Affiliated Hospital of Zhengzhou University.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article: .

References

Alzheimer’s Association (2021) 2021 Alzheimer’s disease facts and figures. Alzheimers Dement 17, 327–406.

Prince

, Wimo

, Guerchet

, Ali

G-C

, Wu

Y-T

, Prina

(2015) World Alzheimer Report 2015. The Global Impact of Dementia: An Analysis of Prevalence, Incidence, Cost and Trends. Alzheimer’s Disease International,–London.

Lista

, O’Bryant

, Blennow

, Dubois

, Hugon

, Zetterberg

, Hampel

(2015) Biomarkers in sporadic and familial Alzheimer’s disease. J Alzheimers Dis 47, 291–317.

Richard

(2019) Risk factors for Alzheimer’s disease. Folia Neuropathol 57, 87–105.

Sochocka

, Zwolinska

, Leszek

(2017) The infectious etiology of Alzheimer’s disease. Curr Neuropharmacol 15, 996–1009.

Chen

, Wu

, Chang

(2017) Association between chronic periodontitis and the risk of Alzheimer’s disease: A retrospective, population-based, matched-cohort study. Alzheimers Res Ther 9, 56.

Martande

, Pradeep

, Singh

, Kumari

, Suke

, Raju

, Naik

, Singh

, Guruprasad

, Chatterji

(2014) Periodontal health condition in patients with Alzheimer’s disease. Am J Alzheimers Dis Other Demen 29, 498–502.

Ide

, Harris

, Stevens

, Sussams

, Hopkins

, Culliford

, Fuller

, Ibbett

, Raybould

, Thomas

, Puenter

, Teeling

, Perry

, Holmes

(2016) Periodontitis and cognitive decline in Alzheimer’s disease. PLoS One 11, e0151081.

Tzeng

, Chung

, Yeh

, Huang

, Yuh

, Huang

, Lu

, Chang

, Kao

, Chiang

, Chou

, Chien

(2016) Are chronic periodontitis and gingivitis associated with dementia? A nationwide, retrospective, matched-cohort study in Taiwan. Neuroepidemiology 47, 82–93.

10.

Kanagasingam

, Chukkapalli

, Welbury

, Singhrao

(2020) Porphyromonas gingivalis is a strong risk factor for Alzheimer’s disease, J Alzheimers Dis Rep 4, 501–511.

11.

, Tang

, Lu

, Liu

, He

, Jiang

, Zhang

, Wu

(2021) Oral Treponema denticolainfection induces Aβ(1-40) and Aβ(1-42) accumulation in the hippocampus of C57BL/6 mice, J Mol Neurosci 71, 1506–1514.

12.

Diaz-Zuniga

, Munoz

, Melgar-Rodriguez

, More

, Bruna

, Lobos

, Monasterio

, Vernal

, Paula-Lima

(2019) Serotype b of Aggregatibacter actinomycetemcomitans triggers pro-inflammatory responses and amyloid beta secretion in hippocampal cells: A novel link between periodontitis and Alzheimer s disease? J Oral Microbiol 11, 1586423.

13.

Han

, Choi

, Lee

, Park

, Hong

, Lee

(2019) Extracellular RNAs in periodontopathogenic outer membrane vesicles promote TNF-alpha production in human macrophages and cross the blood-brain barrier in mice. FASEB J 33, 13412–13422.

14.

Escapa

, Chen

, Huang

, Gajare

, Dewhirst

, Lemon

(2018) New insights into human nostril microbiome from the expanded Human Oral Microbiome Database (eHOMD): A resource for the microbiome of the human aerodigestive tract. mSystems 3, e00187–e00118.

15.

Curtis

, Diaz

, Van Dyke

(2020) The role of the microbiota in periodontal disease. Periodontol 2000 83, 14–25.

16.

Verma

, Garg

, Dubey

(2018) Insights into the human oral microbiome. Arch Microbiol 200, 525–540.

17.

Lievense

, Bierma-Zeinstra

, Verhagen

, Verhaar

, Koes

(2004) Influence of hip dysplasia on the development of osteoarthritis of the hip. Ann Rheum Dis 63, 621–626.

18.

McHugh

(2012) Interrater reliability: The kappa statistic. Biochem Med (Zagreb) 22, 276–282.

19.

Riviere

, Riviere

, Smith

(2002) Molecular and immunological evidence of oral Treponema in the human brain and their association with Alzheimer’s disease. , Oral Microbiol Immunol 17, 113–118.

20.

Kamer

, Craig

, Pirraglia

, Dasanayake

, Norman

, Boylan

, Nehorayoff

, Glodzik

, Brys

, de Leon

(2009) TNF-alpha and antibodies to periodontal bacteria discriminate between Alzheimer’s disease patients and normal subjects. J Neuroimmunol 216, 92–97.

21.

Sparks Stein

, Steffen

, Smith

, Jicha

, Ebersole

, Abner

, Dawson

3rd (2012) Serum antibodies to periodontal pathogens are a risk factor for Alzheimer’s disease. Alzheimers Dement 8, 196–203.

22.

Noble

, Scarmeas

, Celenti

, Elkind

, Wright

, Schupf

, Papapanou

(2014) Serum IgG antibody levels to periodontal microbiota are associated with incident Alzheimer disease, PLoS One 9, e114959.

23.

Dominy

, Lynch

, Ermini

, Benedyk

, Marczyk

, Konradi

, Nguyen

, Haditsch

, Raha

, Griffin

, Holsinger

, Arastu-Kapur

, Kaba

, Lee

, Ryder

, Potempa

, Mydel

, Hellvard

, Adamowicz

, Hasturk

, Walker

, Reynolds

, Faull

RLM

, Curtis

, Dragunow

, Potempa

(2019) Porphyromonas gingivalis in Alzheimer’s disease brains: Evidence for disease causation and treatment with small-molecule inhibitors, Sci Adv 5, eaau3333.

24.

Beydoun

, Beydoun

, Hossain

, El-Hajj

, Weiss

, Zonderman

(2020) Clinical and bacterial markers of periodontitis and their association with incident all-cause and Alzheimer’s disease dementia in a large national survey. J Alzheimers Dis 75, 157–172.

25.

Cirstea

, Kliger

, MacLellan

, Yu

, Langlois

, Fan

, Boroomand

, Kharazyan

, Hsiung

RGY

, MacVicar

, Chertkow

, Whitehead

, Finlay

, Appel-Cresswell

(2022) The oral and fecal microbiota in a Canadian cohort of Alzheimer’s disease. J Alzheimers Dis 87, 247–258.

26.

Poole

, Singhrao

, Kesavalu

, Curtis

, Crean

(2013) Determining the presence of periodontopathic virulence factors in short-term postmortem Alzheimer’s disease brain tissue. J Alzheimers Dis 36, 665–677.

27.

Bathini

, Foucras

, Dupanloup

, Imeri

, Perna

, Berruex

, Doucey

, Annoni

, Auber Alberi

(2020) Classifying dementia progression using microbial profiling of saliva. Alzheimers Dement (Amst) 12, e12000.

28.

Leblhuber

, Huemer

, Steiner

, Gostner

, Fuchs

(2020) Knock-on effect of periodontitis to the pathogenesis of Alzheimer’s disease? Wien Klin Wochenschr 132, 493–498.

29.

Panzarella

, Mauceri

, Baschi

, Maniscalco

, Campisi

, Monastero

(2020) Oral health status in subjects with amnestic mild cognitive impairment and Alzheimer’s disease: Data from the Zabút Aging Project. J Alzheimers Dis 87, 173–183.

30.

Holmer

, Aho

, Eriksdotter

, Paulin

, Pietiainen

, Auvinen

, Schultzberg

, Pussinen

, Buhlin

(2021) Subgingival microbiota in a population with and without cognitive dysfunction. J Oral Microbiol 13, 1854552.

31.

Liu

, Jiao

, Liao

, Guo

, Yuan

, Wang

, Xiao

, Zhang

, Tang

, Shen

(2019) Analysis of salivary microbiome in patients with Alzheimer’s disease. J Alzheimers Dis 72, 633–640.

32.

, Lee

, Salamanca

, Yao

, Su

, Wang

, Hu

, Chang

(2021) Oral microbiota changes in elderly patients, an indicator of Alzheimer’s disease. Int J Environ Res Public Health 18, 4211.

33.

Guo

, Li

, Yao

, Liu

, Chen

, Zhou

, Ji

, Zeng

, Du

(2021) Profiling the oral microbiomes in patients with Alzheimer’s disease. Oral Dis, doi: 10.1111/odi.14110.

34.

Gupta

, Paul

, Dutta

(2017) Geography, ethnicity or subsistence-specific variations in human microbiome composition and diversity. Front Microbiol 8, 1162.

35.

Silva

, Cascales

(2021) Molecular strategies underlying Porphyromonas gingivalis virulence. J Mol Biol 433,, 166836.

36.

Gui

, Dashper

, Slakeski

, Chen

, Reynolds

(2016) Spheres of influence: Porphyromonas gingivalis outer membrane vesicles. Mol Oral Microbiol 31, 365–378.

37.

Poole

, Singhrao

, Chukkapalli

, Rivera

, Velsko

, Kesavalu

, Crean

(2015) Active invasion of Porphyromonas gingivalis and infection-induced complement activation in ApoE-/- mice brains, J Alzheimers Dis 43, 67–80.

38.

Ishida

, Ishihara

, Ishida

, Tada

, Funaki-Kato

, Hagiwara

, Ferdous

, Abdullah

, Mitani

, Michikawa

, Matsushita

(2017) Periodontitis induced by bacterial infection exacerbates features of Alzheimer’s disease in transgenic mice. NPJ Aging Mech Dis 3, 15.

39.

Singhrao

, Chukkapalli

, Poole

, Velsko

, Crean

, Kesavalu

(2017) Chronic Porphyromonas gingivalis infection accelerates the occurrence of age-related granules in ApoE(-)(/)(-) mice brains, J Oral Microbiol 9, 1270602.

40.

Ding

, Ren

, Yu

, Zhou

(2018) Porphyromonas gingivalis, a periodontitis causing bacterium, induces memory impairment and age-dependent neuroinflammation in mice. Immun Ageing 15, 6.

41.

Ilievski

, Zuchowska

, Green

, Toth

, Ragozzino

, Le

, Aljewari

, O’Brien-Simpson

, Reynolds

, Watanabe

(2018) Chronic oral application of a periodontal pathogen results in brain inflammation, neurodegeneration and amyloid beta production in wild type mice, PLoS One 13, e0204941.

42.

Cheng

, Tang

, Liang

, Zeng

, Liu

, Lian

, Wu

(2020) The effects of Porphyromonas gingivalis on the apoptosis of hippocampal cells in Sprague-Dawley rats and its underlying mechanisms. Int J Clin Exp Med 13, 300–309.

43.

Haditsch

, Roth

, Rodriguez

, Hancock

, Cecere

, Nguyen

, Arastu-Kapur

, Broce

, Raha

, Lynch

, Holsinger

, Dominy

, Ermini

(2020) Alzheimer’s disease-like neurodegeneration in Porphyromonas gingivalis infected neurons with persistent expression of active gingipains. J Alzheimers Dis 75, 1361–1376.

44.

Yang

, Zhou

(2019) Neuroinflammation in the central nervous system: Symphony of glial cells. Glia 67, 1017–1035.

45.

Qiu

, Yuan

, He

, Chen

, Liao

, Li

, Zhou

, Song

(2021) Lipopolysaccharide preparation derived from Porphyromonas gingivalis induces a weaker immuno-inflammatory response in BV-2 microglial cells than Escherichia coli by differentially activating TLR2/4-mediated NF-kappa B/STAT3 signaling pathways, Front Cell Infect Microbiol 11, 606986.

46.

Nativel

, Couret

, Giraud

, Meilhac

, d’Hellencourt

, Viranaicken

, Da Silva

(2017) Porphyromonas gingivalis lipopolysaccharides act exclusively through TLR4 with a resilience between mouse and human. Sci Rep 7, 15789.

47.

Memedovski

, Czerwonka

, Han

, Mayer

, Luce

, Klemm

, Hall

, Mayer

AMS

(2020) Classical and alternative activation of rat microglia treated with ultrapure Porphyromonas gingivalis lipopolysaccharide in vitro. Toxins (Basel) 12, 333.

48.

Tran

VTA

, Kang

, Kim

, Cho

(2021) Oral pathogenic bacteria-inducing neurodegenerative microgliosis in human neural cell platform. Int J Mol Sci 22, 6925.

49.

Sato

, Matsumoto

, Kawaguchi

, Seya

, Matsumiya

, Ding

, Aizawa

, Imaizumi

(2021) Porphyromonas gingivalis lipopolysaccharide induces interleukin-6 and c-c motif chemokine ligand 2 expression in cultured hCMEC/D3 human brain microvascular endothelial cells. Gerodontology 39, 139–147.

50.

Tetz

, Pinho

, Pritzkow

, Mendez

, Soto

, Tetz

(2020) Bacterial DNA promotes tau aggregation. Sci Rep 10, 2369.

51.

Tetz

, Tetz

(2021) Bacterial extracellular DNA promotes beta-amyloid aggregation. Microorganisms 9, 1301.

52.

Byrne

, Dashper

, Darby

, Adams

, Hoffmann

, Reynolds

(2009) Progression of chronic periodontitis can be predicted by the levels of Porphyromonas gingivalis and Treponema denticola in subgingival plaque. Oral Microbiol Immunol 24, 469–477.

53.

Orth

, O’Brien-Simpson

, Dashper

, Reynolds

(2011) Synergistic virulence of Porphyromonas gingivalis and Treponema denticola in a murine periodontitis model, Mol Oral Microbiol 26, 229–240.

54.

Zhu

, Dashper

, Chen

, Crawford

, Slakeski

, Reynolds

(2013) Porphyromonas gingivalis and Treponema denticola synergistic polymicrobial biofilm development, PLoS One 8, e71727.

55.

Tan

, Seers

, Dashper

, Mitchell

, Pyke

, Meuric

, Slakeski

, Cleal

, Chambers

, McConville

, Reynolds

(2014) Porphyromonas gingivalis and Treponema denticola exhibit metabolic symbioses, PLoS Pathog 10, e1003955.

56.

, Slakeski

, Butler

, Veith

, Chen

, Liu

, Hoffmann

, Dashper

, Reynolds

(2019) The role of Treponema denticolamotility in synergistic biofilm formation with Porphyromonas gingivalis. Front Cell Infect Microbiol 9, 432.

57.

, Kin

, Dashper

, Slakeski

, Butler

, Reynolds

(2016) Bacterial interactions in pathogenic subgingival plaque. Microb Pathog 94, 60–69.

58.

Kin

, Butler

, Slakeski

, Hoffmann

, Dashper

, Reynolds

(2020) Metabolic cooperativity between Porphyromonas gingivalis and Treponema denticola. J Oral Microbiol 12, 1808750.

59.

Van Dyke

, Bartold

, Reynolds

(2020) The nexus between periodontal inflammation and dysbiosis. Front Immunol 11, 511.

60.

Singhrao

, Harding

, Simmons

, Robinson

, Kesavalu

, Crean

(2014) Oral inflammation, tooth loss, risk factors, and association with progression of Alzheimer’s disease. J Alzheimers Dis 42, 723–737.

61.

Singhrao

, Harding

(2020) Is Alzheimer’s disease a polymicrobial host microbiome dysbiosis? Expert Rev Anti Infect Ther 18, 275–277.

62.

Noble

, Borrell

, Papapanou

, Elkind

, Scarmeas

, Wright

(2009) Periodontitis is associated with cognitive impairment among older adults: Analysis of NHANES-III. J Neurol Neurosurg Psychiatry 80, 1206–1211.

63.

Villemagne

, Burnham

, Bourgeat

, Brown

, Ellis

, Salvado

, Szoeke

, Macaulay

, Martins

, Maruff

, Ames

, Rowe

, Masters

, Australian Imaging Biomarkers and Lifestyle (AIBL) Research Group (2013) Amyloid beta deposition, neurodegeneration, and cognitive decline in sporadic Alzheimer’s disease: A prospective cohort study. Lancet Neurol 12, 357–367.

64.

Mager

, Ximenez-Fyvie

, Haffajee

, Socransky

(2003) Distribution of selected bacterial species on intraoral surfaces. J Clin Periodontol 30, 644–654.

65.

Hung

, Chang

, Huang

, Nouchi

, Cheng

(2022) Gut microbiota in patients with Alzheimer’s disease spectrum: A systematic review and meta-analysis. Aging (Albany NY) 14, 477–496.

66.

Shen

, Guan

, Zhang

, Yuan

, Tan

, Zhai

, Hao

, Gu

, Han

(2020) New mechanism of neuroinflammation in Alzheimer’s disease: The activation of NLRP3 inflammasome mediated by gut microbiota. Prog Neuropsychopharmacol Biol Psychiatry 100, 109884.

67.

Chen

, Ahn

, Kang

, Liu

, Alam

, Ye

(2020) Gut dysbiosis contributes to amyloid pathology, associated with C/EBPbeta/AEP signaling activation in Alzheimer’s disease mouse model, Sci Adv 6, eaba0466.

68.

Park

, Hwang

, Lim

, Ok

, Lee

, Chun

, Park

, Hu

, Chung

, Song

(2021) Oral-gut microbiome axis in gastrointestinal disease and cancer. Cancers (Basel) 13, 2124.