Abstract Background Periodontitis and atherosclerosis are both chronic inflammatory diseases with a high prevalence. Increasing evidence supports the independent association between severe periodontitis and atherosclerotic cardiovascular disease, in which oral microorganisms may play an important role. We aimed to evaluate the characteristic changes of salivary microbiome and metabolome in patients with carotid atherosclerosis (CAS) and periodontitis. Methods and Results The subjects were obtained from a cross‐sectional study that included 1933 participants aged 40 years or older from rural northeast China. The study enrolled 48 subjects with CAS and 48 controls without CAS matched by sex, age, body mass index, and prevalence of hypertension, diabetes, and dyslipidemia. We performed full‐length 16S rDNA gene sequencing and untargeted metabolomics of saliva samples from 96 subjects. We found that CAS was closely associated with an increased abundance of Streptococcus, Lactobacillus, and Cutibacterium. Furthermore, patients with CAS had higher prevalence of severe periodontitis than the control group. Notably, periodontal pathogens such as Tannerella and Anaeroglobus were not only associated with periodontitis but also enriched in patients with CAS, whereas periodontal health‐associated Neisseria was more abundant in those without CAS. We also identified 2 lipid metabolism pathways, including glycerophospholipid and sphingolipid metabolism, as associated with CAS. The levels of trimethylamine N‐oxide and inflammatory mediator leukotriene D4 were significantly higher in patients with CAS, whereas the levels of carnosine were significantly lower, than those in controls. Additionally, serum levels of inflammatory marker high‐sensitivity C‐reactive protein were significantly increased in CAS and positively correlated with the abundance of Anaeroglobus and leukotriene D4 in saliva. Conclusions Our study suggests that characteristic changes in salivary microbiota and metabolites are closely related to CAS, and periodontitis and associated microorganisms may be involved in the initiation and progression of CAS. Keywords: atherosclerosis, metabolomic, microbiome, saliva, severe periodontitis Subject Categories: Atherosclerosis __________________________________________________________________ Nonstandard Abbreviations and Acronyms CAS carotid atherosclerosis FDR false discovery rate IMT intima‐media thickness LTD4 leukotriene D4 OPLS‐DA orthogonal least partial squares discriminant analysis TMAO trimethylamine N‐oxide VIP variable importance in the projection Clinical Perspective. What Is New? * The subjects were obtained from a cross‐sectional study that included 1933 participants from rural areas in northeast China; this study evaluated the alterations in the salivary microbiome and metabolome in 48 subjects with carotid atherosclerosis and 48 controls without carotid atherosclerosis matched by cardiovascular risk factors. * Severe periodontitis and associated microorganisms are strongly related to carotid atherosclerosis; serum levels of inflammatory marker high‐sensitivity C‐reactive protein were significantly increased in carotid atherosclerosis and positively correlated with the abundance of Anaeroglobus and inflammatory mediator leukotriene D4 in saliva. What Are the Clinical Implications? * It is imperative to manage the risk factors associated with cardiovascular disease in rural northeast China. Notably, we should pay more attention to the associations between periodontitis and atherosclerotic cardiovascular disease. * Risk screening by detecting related microorganisms and dietary carnosine supplementation may hold considerable importance in the prevention and management of cardiovascular disease. Cardiovascular disease, which includes heart and blood vessel diseases such as coronary heart disease and stroke, is the leading cause of death in humans. In China, more than 4 million people die from cardiovascular disease every year.[40] ^1 The Chinese rural population, comprising the majority of the country's inhabitants, commonly experiences low income levels and educational attainment, as well as limited access to health care services, resulting in a significant burden of cardiovascular disease.[41] ^2 Atherosclerosis is the pathological basis for the occurrence of cardiovascular diseases and considered to be an inflammatory disease, whose main characteristics are vascular inflammation and subintima lipid accumulation.[42] ^3 Carotid atherosclerosis (CAS) reflects the formation and evolution of systemic atherosclerotic disease and is significantly associated with cardiovascular disease.[43] ^4 Increased carotid artery intima media thickness (IMT) and carotid plaque formation are the main indicators of CAS evaluated by carotid artery ultrasonography, as well as important predictors of cardiovascular risk. Gut microbiota, such as Streptococcus, Escherichia, and Lactobacillus, and gut microbe‐derived metabolites including trimethylamine N‐oxide (TMAO) and short‐chain fatty acids, have been shown to be involved in atherosclerosis, thereby playing an important role in cardiovascular disease.[44] ^5 , [45]^6 , [46]^7 , [47]^8 However, there are relatively few studies on oral microbiota in patients with atherosclerosis. The oral cavity is a major gateway to the human body and contains the second most diverse microbiota following the gut.[48] ^9 Periodontitis is an inflammatory disease of oral cavity primarily caused by dysbacteriosis, characterized by the destruction of periodontal tissues.[49] ^10 In addition, growing evidence reveals that oral bacteria can promote the occurrence and development of systemic diseases by spreading through the body through hematogenous and enteral routes or causing systemic inflammation and autoimmune reactions.[50] ^9 , [51]^11 The presence of a variety of oral bacteria in human atherosclerosis plaques suggests that oral bacteria may be an important factor involved in the formation and development of atherosclerosis.[52] ^12 Both epidemiological and animal studies have provided strong evidence showing that oral microbiota might play a significant role in atherosclerosis. The abundance of the periodontal pathogen Aggregatibacter actinomycetemcomitans was shown to be increased in saliva of patients with both symptomatic and asymptomatic coronary disease compared with healthy controls.[53] ^13 With respect to metabolomics, the metabolites generated by oral microbiota, including indole, nitrate, and nitrite, play an important role in the development of atherosclerosis.[54] ^14 , [55]^15 Saliva is secreted from the salivary glands and has a complex composition, containing organic molecules such as lysozymes, mucin, fatty acid and amino acid, and some inorganic substances such as Na^+, K^+, and Ca^2+. It also contains a large number of microorganisms that are associated with oral and systemic diseases. Saliva has multiple functions, such as mouth cleaning, acting as an antimicrobial, and aiding digestion. Salivary microorganisms are closely related to subgingival periodontal pathogens, and the translocation of oral pathogens resulting from periodontitis by swallowing can induce systemic inflammation linked to intestinal dysbiosis.[56] ^16 , [57]^17 In addition, saliva assists in reflecting the body condition because biomarkers in the blood circulation can be secreted into the saliva by infiltrating into acinar cells.[58] ^18 Consequently, multiomic analysis of salivary microbiome and metabolome may reveal the mechanistic links between oral microbiota and atherosclerosis. An increasing number of studies have demonstrated a potential correlation between alterations in oral microbiota and metabolites and the progression of atherosclerosis. However, the specific details of these alterations have not been thoroughly investigated. The objective of this study was to identify distinctive alterations in the salivary microbiota and metabolites that could potentially contribute to the development of CAS and to evaluate potential associations between the oral microbiome, metabolome, and clinical parameters in CAS by microbiome and metabolomic analysis in subjects from rural northeast China. Methods Data Sharing The microbiome and metabolomics raw data have been made publicly available at the database of the National Center for Biotechnology Information Gene Expression Omnibus and MetaboLights respectively and can be accessed at [59]https://www.ncbi.nlm.nih.gov/geo/ and [60]https://www.ebi.ac.uk/metabolights/. Subjects Recruitment This study is based on a cross‐sectional survey conducted in rural areas of northeast China in April 2021. Fifteen villages were randomly selected from Chaoyang County, Liaoning Province. All permanent residents aged 40 years and over were eligible, and those who were pregnant or had a mental disorder were excluded. We enrolled 1933 subjects. The demographic information, cardiovascular risk factors, and oral health behaviors were collected using a questionnaire. All participants went through measurement of height, weight, and blood pressure. Experienced sonographers performed a carotid artery ultrasonography in each patient. Individuals who had carotid intima‐media thickening (IMT ≥1.0 mm) or at least 1 carotid artery plaque were diagnosed with CAS by ultrasound examination. Carotid plaque presence was defined as a focal absolute wall thickness (IMT >1.5 mm) or a relative focal thickening of >50% of the adjacent IMT.[61] ^19 These patients with CAS comprised the case group. Subjects with negative ultrasound results were assigned to the control group. Subjects who missed carotid artery ultrasonography and periodontal examination, had <10 remaining teeth, smoked currently, developed immune‐inflammatory diseases, underwent periodontal treatment within 6 months, used antibiotics within 3 months, received treatment for cardiovascular diseases, and took lipid‐regulating drugs, antihypertensive drugs, or hypoglycemic drugs within 1 month were excluded from further analysis. Case and control subjects were paired according to sex, age (using a 3‐year threshold of tolerance), body mass index (using 1‐value as the threshold of tolerance), and prevalence of hypertension, diabetes, and dyslipidemia. Subjects who could not be paired were excluded from the study. Therefore, 96 subjects were included in the final analysis (Figure [62]1). All participants provided informed consent and the study was approved by the Medical Ethical Review Committee of Affiliated Stomatology Hospital of China Medical University. Figure 1. Flow chart of population selection. Figure 1 [63]Open in a new tab BMI indicates body mass index. Sample Collection and Oral Examination Peripheral blood and saliva samples were collected after fasting for 8 hours. Total cholesterol, triglyceride, high‐density lipoprotein‐cholesterol, low‐density lipoprotein‐cholesterol, fasting plasma glucose, hemoglobin A1c were measured by biochemical analyzers in blood samples. Serum concentrations of hs‐CRP (high‐sensitivity C‐reactive protein; Cusabio, Wuhan, China), IL‐1β (interleukin‐1β; Thermo Fisher Scientific, MA, USA), and IL‐6 (Multisciences, Hangzhou, China) were quantified with ELISA. In order to collect saliva samples, participants were not allowed to eat breakfast, drink water, brush their teeth, or use any mouthwash. A 15 mL tube was used for collecting 2 to 5 mL of saliva without any stimulation.[64] ^20 Collected samples were immediately placed on dry ice and stored at −80 °C within 4 hours. Dentists conducted a complete mouth examination on all participants. Diagnosis of periodontitis was made based on the 2017 World Workshop on the Classification of Periodontal and Peri‐Implant Diseases and Conditions.[65] ^21 The 96 volunteers were further divided into severe periodontitis with CAS (A group, n= 45), severe periodontitis without CAS (C2 group, n=28), gingivitis or mild to moderate periodontitis without CAS (C1 group, n=20), and gingivitis or mild to moderate periodontitis with CAS (n=3) according to periodontal status.[66] ^22 DNA Extraction, Polymerase Chain Reaction Amplification, and 16S rDNA Sequencing Bacterial DNA from saliva samples was extracted using the DNeasy PowerSoil Pro Kit (QIAGEN, Hilden, Germany). Polymerase chain reaction was performed using bacterial primers 27F (5′‐AGRGTTYGATYMTGGCTCAG‐3′) and 1492R (5′‐RGYTACCTTGTTACGACTT‐3′) to amplify the bacterial 16S rRNA genes.[67] ^23 AMPure PB beads (Pacific Biosciences, Menlo Park, CA, USA) and QuantusTM Fluorometers (Promega, WI, USA) were used to purify and quantify polymerase chain reaction products. DNA library was constructed using the SMRTbell prep kit 3.0 (Pacific Biosciences, Menlo Park, CA, USA) and then applied to single‐end sequencing on the PacBio Sequel IIe System (Pacific Biosciences, Menlo Park, CA, USA). Reads were barcode identified and length filtered, and sequences with a length <1000 or >1800 bp were removed. Operational taxonomic units were created from optimized sequences at the cutoff of 97% using UPARSE (version 7.1).[68] ^24 The sequences of all samples were rarefied to the minimum sequence to reduce the effects of sequencing depth. The rarefaction curves and alpha diversity indices, including Shannon index and Chao index, were calculated with Mothur (version 1.30.1).[69] ^25 Differences in Shannon index and Chao index between the 2 groups were evaluated using the Wilcoxon rank‐sum test, and differences among multiple groups were evaluated by Kruskal–Wallis test. The differences in microbial community composition were evaluated by principal coordinated analysis based on weighted UniFrac using the Vegan R package. The Adonis test was performed for statistical significance using the Vegan R package. The relative abundance of genera was log10 transformed before the analysis of differences between groups. Differentially abundant genera were identified by Wilcoxon rank‐sum test. Metabolite Extraction and Untargeted Liquid Chromatography–Tandem Mass Spectrometry Analysis The untargeted liquid chromatography–tandem mass spectrometry analysis of saliva samples was carried out on a Thermo UHPLC‐Q Exactive HF‐X system (Thermo Fisher Scientific, MA, USA). First, metabolites were extracted. The quality control samples were composed of 10 μL supernatant from each sample. Progenesis QI (Waters Corporation, Milford, MA, USA) software was applied to preprocess the raw data and export a 3‐dimensional data matrix in CSV format. The data matrix obtained by searching database was analyzed by using the Majorbio cloud platform. First, the data matrix was preprocessed, as follows: Peaks with a zero value in >80% of both sets of samples were removed. After filtering, for specific samples with metabolite levels below the lower limit of quantification, the minimum metabolite value was estimated, and each metabolic signature was normalized to the sum. The sum normalization method was applied to obtain the normalized data matrix. Meanwhile, the variables of quality control samples with relative SD >30% were excluded and log10 logarithmized. Through preprocessing, 880 and 582 metabolites were removed in positive and negative ion modes, respectively. Then, orthogonal least partial squares discriminant analysis (OPLS‐DA) was performed using the R package “ropls” (Version 1.6.2), and the stability of the model was evaluated by 7‐cycle interactive validation. Multiple tests were adjusted using Benjamini and Hochberg false discovery rate (FDR). The metabolites with variable importance in the projection (VIP) >1 and P[fdr] <0.05 (Student's t test) were considered significant. Metabolic enrichment and pathway analysis were performed using the Kyoto Encyclopedia of Genes and Genomes database. Python packages “scipy.stats” and Fisher's exact test were used to perform enrichment analysis to obtain the most relevant biological pathways of disease. Multiomics Correlation Analyses Redundancy analysis was performed to examine the associations between clinical parameters and CAS‐related microbial signatures. Additionally, we calculated Spearman's correlations between significantly altered genera, metabolites, and clinical parameters and visualized them by heatmaps and network diagram using the ggplot2, igraph, and ggraph package in R. Biomarker Identification The receiver operating curve analysis was operated to assess the diagnostic ability of candidate microbiota and metabolites for diagnosis of CAS using R package pROC. Statistical Analysis The statistical analyses were carried out using R version 4.2.1. Normally distributed continuous values are presented using mean±SD, and nonnormally distributed continuous values are presented using median (interquartile range). Pairwise comparisons of continuous variables were conducted using Student's t test when assumptions of normality and homogeneity of variance were met. When normality was present but variance was not homogeneous, Welch's t test was employed. Nonnormal data with heterogeneous variance were compared using the Wilcoxon rank‐sum test. One‐way ANOVA was employed to compare multiple group continuous variables that satisfied the assumptions of normality and homogeneity of variance. In cases where the multiple group continuous variables exhibited normality but not homogeneity of variance, Welch one‐way ANOVA was used. For nonnormal data with inhomogeneity of variance, the Kruskal–Wallis test was used for comparison. Categorical variables are expressed as total numbers (percentage). Categorical variables comparisons were performed using chi‐square test or chi‐square test with Yates' continuity corrected. Significant difference was defined as a P value <0.05. Two sensitivity analyses were performed. First, we reanalyzed the differences of several differential genera and metabolites between CAS and controls by excluding participants with diabetes (n=34) or dyslipidemia (n=32). Second, considering that CAS is related to sex and age, we stratified participants according to sex and age (≤60 or >60 years) to examine whether the results vary in different subgroups. The relative abundance of genera was log10 transformed. Results Characteristics of the Participants The demographic and clinical characteristics of patients with CAS and controls without CAS are presented in Table [70]S1. CAS and controls were paired according to sex, age, body mass index, and prevalence of hypertension, diabetes, and dyslipidemia. After matching, there were no statistically significant differences in blood lipid levels, blood glucose levels, and blood pressure between the 2 groups. Additionally, there were no significant disparities in dietary habits and frequency of tooth brushing between the 2 groups. All subjects did not use interdental cleaning devices (either dental floss or interdental brush) or any mouthwash. However, it was observed that patients with CAS exhibited greater IMT thickness and a higher prevalence of severe periodontitis (P<0.001) compared with controls. Compared with the control group, the serum hs‐CRP concentration in the group with CAS was significantly increased (P<0.05), whereas there was no significant difference in serum IL‐1β and IL‐6 concentrations between the 2 groups (P>0.05) (Figure [71]S1). Increased Alpha Diversity and Altered Microbial Composition in CAS A rarefaction curve based on the Sobs index suggested that sequencing depth was adequate (Figure [72]S2A). In terms of alpha diversity, the Shannon index and Chao index, which reflects the community richness and diversity respectively, were significantly higher in the group with CAS than the control group (P<0.05) (Figure [73]2A and [74]2B). In order to assess the variation in microbial community composition between the 2 groups, beta diversity analysis was conducted. The separation shown in the weighted UniFrac principal coordinated analysis diagram was apparent and Adonis revealed significant differences (R ^2=0.064, P<0.01) (Figure [75]2C). Figure 2. Salivary microbiome variations in CAS. Figure 2 [76]Open in a new tab A, B, Shannon index and Chao index of patients with CAS and controls. C, Principal coordinates analysis for the group with CAS and control group. D, Bar plot of relative abundance for CAS and controls from genus level. E, Boxplots show the relative abundance of genera altered in CAS compared with the control group analyzed by Wilcoxon rank‐sum test. *P<0.05, **P<0.01, ***P<0.001. CAS indicates carotid atherosclerosis; Con, controls; PC1, principal component 1; and PC2, principal component 2. We subsequently aimed to identify the microbial profiles associated with CAS. The phylum Firmicutes was found to be the most prevalent, constituting 66.19% and 54.96% of the salivary microbiota in the group with CAS and control group, respectively (Figure [77]S2B). At the genus level, Streptococcus exhibited the highest prevalence, with a relative abundance of 44.16% in the group with CAS and 34.30% in the control group (Figure [78]2D). The second most dominant genus was Neisseria, with a prevalence of 22.95% in the control group and 12.83% in the group with CAS (Figure [79]2D). To identify significant taxonomic differences, the Wilcoxon rank‐sum test was employed, revealing 13 genera that exhibited substantial variations in abundance between the 2 groups (P<0.05) (Figure [80]2E). Specifically, Streptococcus (P=0.002), unclassified_o__Lactobacillales (P=0.003), Dialister (P=0.002), Rikenellaceae_RC9_gut_group (P=0.010), Tannerella (P=0.044), Atopobium (P=0.033), Lactobacillus (P=0.008), Butyrivibrio (P=0.030), Shuttleworthia (P=0.020), Cutibacterium (P<0.001), and Anaeroglobus (P=0.020) were enriched in the group with CAS, and Neisseria (P=0.010) and Amnipila (P=0.027) were enriched in the control group. Tannerella was worthy of attention, because species belonging to this genus (Tannerella forsythia [T. forsythia]) is part of the periodontal “red complex” pathogens.[81] ^26 These results showed a remarkable difference in oral microbiota composition between the group with CAS and the controls. Severe Periodontitis May Be Associated With CAS In order to understand whether periodontal status had impacts on salivary microbiota in our study, we further compared the microbial composition stratified by periodontal status. The 96 volunteers were divided into severe periodontitis with CAS (A group, n=45), severe periodontitis without CAS (C2 group, n=28), gingivitis or mild to moderate periodontitis without CAS (C1 group, n=20), and gingivitis or mild to moderate periodontitis with CAS (n=3). As the number of patients with CAS and gingivitis or mild to moderate periodontitis was too small, we did not include them in subsequent analyses. The demographic and clinical characteristics of the subgroups are shown in Table [82]S2. There were no significant differences in the cardiovascular disease risk factors among the groups. However, there were significant differences in the serum levels of hs‐CRP among C1, C2, and A groups, with levels increasing gradually across the groups (P<0.05) (Figure [83]S3). The Chao index (P<0.01) was significantly higher in the A group than in either the C2 group or the C1 group in descending order (Figure [84]3A). Salivary microbiota of the C1, C2, and A groups could distinguish apparently in the weighted UniFrac principal coordinated analysis diagram and Adonis revealed significant differences (R ^2=0.075, P<0.01) (Figure [85]3B). By further comparing the significantly altered genera in the subgroup, the relative abundance of Streptococcus, Lactobacillus, and Cutibacterium was significantly increased in A group compared with C1 and C2 groups (Figure [86]3C). Notably, Tannerella, Anaeroglobus, and Scardovia were found to show increasing trend and Neisseria was found to show decreasing trend from C1 group through C2 group to A group. These genera have been associated with periodontitis in previous studies.[87] ^26 , [88]^27 , [89]^28 , [90]^29 Therefore, this study revealed their potential contribution to periodontitis and atherosclerosis and periodontitis may be a potential risk factor for atherosclerosis. Figure 3. Salivary microbiome variations among 3 subgroups. Figure 3 [91]Open in a new tab A, Chao index of C1, C2, and A groups. B, principal coordinates analysis for C1, C2, and A groups. C, The abundance differentiation of the 13 CAS‐associated genera between 3 subgroups are visualized by box‐plot according to the Wilcoxon rank‐sum test. *P<0.05, **P<0.01. A indicates severe periodontitis with CAS; CAS, carotid atherosclerosis; C1, gingivitis or mild to moderate periodontitis without CAS; C2, severe periodontitis without CAS; PC1, principal component 1; and PC2, principal component 2. Salivary Metabolomic Alterations in CAS We subsequently performed untargeted liquid chromatography–tandem mass spectrometry analysis of salivary samples. Significant variations in the metabolomic profile were observed within the cohort of individuals with CAS. The application of OPLS‐DA analysis revealed a clear distinction between patients with CAS and controls based on their distinct metabolic differences (Figure [92]4A and [93]4B). Furthermore, the permutation test provided evidence that the OPLS‐DA model is valid and that no overfitting exists (Figure [94]S4A and [95]S4B). Finally, a total of 131 metabolites were identified as differentially abundant between patients with CAS and controls. Among them, the levels of TMAO, lysophosphatidyl choline (15:0/0:0), leukotriene D4 (LTD4), and 35 other metabolites were significantly higher in the group with CAS compared with the control group. Conversely, the levels of lathosterol, sphinganine, carnosine, and 96 other metabolites demonstrated a significant decrease in the group with CAS compared with the control group. The Kyoto Encyclopedia of Genes and Genomes pathway analysis revealed that differentially abundant metabolites were enriched in sphingolipid and glycerophospholipid metabolism (impact>0.1, P[fdr] <0.05) (Figure [96]4C and [97]4D). The VIP values of the top 20 metabolites are shown in Figure [98]4E. These metabolites exhibited a VIP value >2.5, indicating their substantial contribution to the OPLS‐DA model in effectively distinguishing between the salivary metabolomes of patients with CAS and controls. Figure 4. Salivary metabolome changes in CAS. Figure 4 [99]Open in a new tab A, B, Orthogonal least partial squares discriminant analysis for the group with CAS and control group. C, D, Metabolomic pathway enrichment analysis using significantly altered metabolites between CAS and controls. E, VIP plot of 20 significantly altered metabolites between CAS and controls. CAS indicates carotid atherosclerosis; Con, control group; DG, diacylglycerol; PA, phosphatidic acid; PC, phosphatidyl choline; PE, phosphatidyl ethanolamine; and VIP, variable importance in the projection. Next, we further explored the changes in metabolites and metabolic pathways associated with CAS in patients with severe periodontitis. A total of 86 differentially abundant metabolites (VIP>1, P [fdr ]<0.05) were identified (Figure [100]S5A and [101]S5B) and enriched in glycerophospholipid and sphingolipid metabolism (Figure [102]S5C). Likewise, the abundances of lathosterol, sphinganine, and carnosine were depleted in CAS compared with those without CAS (Figure [103]S5D). Notably, we found that the saliva levels of the inflammatory mediator LTD4 in A group was significantly higher than that in C1 and C2 groups, and the levels increased gradually in C1, C2, and A groups. Multiomics Analysis Revealed Microbiota–Metabolite Interactions of CAS We conducted Spearman's correlation between the 13 differential genera and the top 50 differential metabolites. As shown in Figure [104]5A, Streptococcus, Tannerella, and Neisseria had the most significant relationships with metabolites. Notably, LTD4 was remarkably increased in CAS and positively associated with CAS‐increased taxa such as Streptococcus, Tannerella, and Anaeroglobus (P<0.05). Taken together, the strong link between differentially abundant microbes and metabolites may confirm the interactions between oral microbes and metabolites in CAS. Figure 5. Correlations between microbiota, metabolites, and clinical parameters (Spearman's correlation analysis). Figure 5 [105]Open in a new tab A, Associations between differential microbiota and top 50 differential metabolites were by correlation heatmap. B, Associations between differential microbiota and clinical parameters were by correlation heatmap. C, Associations between top 50 differential metabolites and clinical parameters were by correlation heatmap. Microbiota and metabolites increased or decreased in the group with CAS are highlighted in red and blue, respectively. *P<0.05. BMI indicates body mass index; CAS, carotid atherosclerosis; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HbA1c, hemoglobin A1C; HDL‐C, high‐density lipoprotein cholesterol; hs‐CRP, high‐sensitivity C‐reactive protein; IL, interleukin; IMT, intima‐media thickness; LDL‐C, low‐density lipoprotein cholesterol; LysoPC, lysophosphatidyl choline; PA, phosphatidic acid; PC, phosphatidyl choline; PE, phosphatidyl ethanolamine; PG, phosphoglyceride; PS, phosphatidyl serine; SBP, systolic blood pressure; TC, total cholesterol; and TG, triacylglycerol. Subsequently, we performed redundancy analysis to identify the potential clinical parameters associated with the microbial profiles. Redundancy analysis found that triglyceride, hs‐CRP, fasting plasma glucose, high‐density lipoprotein‐cholesterol, and hemoglobin A1c showed significant associations with salivary microbiome (P<0.001) (Figure [106]S6). To further investigate the potential role of the altered oral microbiota and metabolites in the pathogenesis of CAS, we next tested for their correlations with clinical parameters using Spearman's correlation (Figure [107]5B and [108]5C). Notably, 2 CAS‐enriched genera (Tannerella and Cutibacterium) were positively correlated with IMT, whereas Neisseria, abundant in controls, was negatively correlated with CAS plaque thickness. CAS‐enriched Streptococcus was positively correlated with triglyceride and hemoglobin A1c but negatively correlated with high‐density lipoprotein‐cholesterol. Genera Anaeroglobus and Atopobium, enriched in CAS, were positively correlated with hs‐CRP, and Neisseria enriched in controls was negatively correlated with hs‐CRP. In addition, CAS‐enriched metabolites tended to positively correlate with disease severity, whereas control‐enriched metabolites displayed opposite trends. It is worth noting that there was a significant positive correlation between saliva LTD4 and serum inflammatory markers hs‐CRP, IL‐1β, and IL‐6. Taken together, these results suggest that some differential genera and metabolites were correlated to clinical parameters. We further filtered out genera and metabolites closely associated with indicators of disease severity according to P<0.05 and |ρ|>0.3. As demonstrated in Figure [109]S7, a total of 9 CAS‐associated genera were significantly correlated with 11 metabolites, and these genera and metabolites were further correlated with IMT and atherosclerosis plaque thickness. Collectively, analysis revealed the close associations between microorganisms, metabolites, and disease severity. Disease Identification and Prediction Based on CAS‐Associated Genera and Metabolites In order to explore the potential applicability of salivary microbial and metabolic profiles in predicting diseases, we performed the receiver operating curve analysis to evaluate the diagnostic accuracy of salivary microbiota and metabolites in discriminating CAS and controls. We identified a salivary microbiome signature composed of 9 significantly altered genera that distinguished CAS from the controls with an area under the curve of 0.806, including 7 genera positively correlated with CAS (Streptococcus, Tannerella, Cutibacterium, Anaeroglobus, Lactobacillus, Dialister and Atopobium) and 2 genera negatively correlated with CAS (Neisseria and Amnipila) (Figure [110]6A). In addition, we selected the top 20 metabolites according to VIP values. Metabolites with an area under the curve <0.95 were eliminated. Finally, we obtained 7 candidate biomarkers, including 7 metabolites negatively correlated with CAS (diacylglycerol (14:1(9Z)/14:1(9Z)/0:0), retinal, phosphatidyl ethanolamine (16:1(9Z)/14:0), phosphatidyl ethanolamine (15:0/16:1(9Z)), lathosterol, phosphatidyl choline (18:0/16:1(9Z)), sphinganine). The area under the curve for the combination of the 7 metabolites was 0.994 (Figure [111]6B). However, the incorporation of genera and metabolites features did not enhance the accuracy of classification (area under the curve: 0.974) (Figure [112]6C). Figure 6. Microbiota and metabolites markers for discriminating CAS from controls. Figure 6 [113]Open in a new tab A, Microbial predictive model. Streptococcus, Tannerella, Cutibacterium, Anaeroglobus, Lactobacillus, Dialister, Atopobium, Neisseria, and Amnipila were included. B, Metabolomic predictive model. DG (14:1(9Z)/14:1(9Z)/0:0), retinal, PE (16:1(9Z)/14:0), PE (15:0/16:1(9Z)), lathosterol, PC (18:0/16:1(9Z)), and sphinganine were included. C, Incorporation of genera and metabolites predictive model. The aforementioned 9 genera and 7 metabolites were included. AUC indicates area under the curve; CAS, carotid atherosclerosis; DG, diacylglycerol; PC, phosphatidyl choline; and PE, phosphatidyl ethanolamine. Sensitivity Analyses In the sensitivity analysis that excluded participants with diabetes or dyslipidemia, the differences between CAS and controls did not change substantially. The abundance of genera (Streptococcus, Lactobacillus, and Cutibacterium) and metabolites (TMAO and LTD4) was significantly higher in the CAS group compared with the control group, and the abundance of carnosine in CAS was significantly lower (Table [114]S3). In addition, the results of stratification analyses according to sex were consistent with these results. However, in the age‐stratified analysis, the abundance of Lactobacillus in the participants ≤60 years and the abundance of TMAO in the participants >60 years showed no significant difference between CAS and controls (Table [115]S4). Discussion The available evidence suggests that the oral microbiota and metabolites have a significant impact on atherosclerosis. This study conducted comprehensive multiomics analyses of the oral microbiome and metabolome in patients with CAS, leading to the identification of disease‐related markers and offering new insights into the underlying mechanisms of the disease (Figure [116]7). Figure 7. CAS is associated with alterations in salivary microbiome and metabolism. Figure 7 [117]Open in a new tab Microbiome and metabolomics analysis of saliva samples from 48 patients with CAS and 48 controls without CAS matched for cardiovascular disease risk factors from rural areas in northeast China revealed that some oral microorganisms and metabolites may be closely related to CAS. CAS indicates carotid atherosclerosis; LTD4, leukotriene D4; TMAO, trimethylamine‐N‐oxide; and UHPLC–MS/MS analysis, untargeted liquid chromatography–tandem mass spectrometry analysis. The subjects were obtained from a cross‐sectional study that included 1933 participants aged 40 years or older from rural areas in northeast China. A total of 48 subjects with CAS and 48 controls without CAS matched by cardiovascular risk factors were enrolled in this study. We found that the prevalence of severe periodontitis in patients with CAS was significantly higher than that in controls (93.8% versus 58.3%, P<0.001). Therefore, we speculate that severe periodontitis may be closely related to the occurrence of CAS. The abundances of several bacterial genera, including Streptococcus, Cutibacterium, and Lactobacillus, increased in patients with CAS irrespective of the presence of severe periodontitis. Consistent with our study, Streptococcus and Cutibacterium resided in the atherosclerotic plaques and Streptococcus and Lactobacillus were found to increase in the oral cavity and gut of patients with atherosclerosis in previous studies.[118] ^5 , [119]^30 , [120]^31 , [121]^32 , [122]^33 Long ago it was realized that Streptococcus residing in the oral cavity can eventually gain access to the bloodstream and cause infective endocarditis.[123] ^34 Streptococcus receptor polysaccharides may stimulate aortic endothelial cell, and cytokines (IL‐6, IL‐8, and monocyte chemoattractant protein‐1) and intercellular adhesion molecule‐1 showed increased expression, thus contributing to cardiovascular disease progression and arterial thrombosis.[124] ^35 Cutibacterium acnes (C. acnes), the most‐studied species of Cutibacterium, is a gram‐positive aerotolerant anaerobe of the normal flora of the skin, oral cavity, large intestine, and urinary tract.[125] ^36 Accumulating evidence reveals that C. acnes could become an opportunistic pathogen involved in various infections.[126] ^37 C. acnes can produce several virulence factors such as Christie–Atkins–Munch‐Peterson factors, leading to a chronic inflammatory reaction and initiating atherogenesis. However, Lactobacillus has been extensively studied for its potential as a probiotic with cholesterol‐lowering and inflammation‐reducing effects in atherosclerosis. The exact role of Lactobacillus in the development of atherosclerosis remains controversial, necessitating further investigation.[127] ^38 , [128]^39 In sum, these results provide insight into oral microbiota characteristics of atherosclerosis and uncover the underlying mechanism of associations between oral microbiota and atherosclerosis. Moreover, it is worth noting that oral bacterial genera such as Tannerella and Anaeroglobus, with higher detection rates in severe periodontitis, also had a higher abundance in CAS, suggesting their potential contribution to severe periodontitis and CAS. T. forsythia is a gram‐negative anaerobe belonging to the “red complex,” which are closely implicated in the development of periodontal disease. Studies have described that T. forsythia can invade and survive within the host cell cytoplasm and thus enter the bloodstream and disseminate to distant sites.[129] ^40 Rangé et al. observed a higher prevalence of T. forsythia in more vulnerable, hemorrhagic plaques and a potential involvement of T. forsythia in neutrophil activation within hemorrhagic atherosclerotic carotid plaques.[130] ^41 Anaeroglobus, also an anaerobic gram‐negative genus, has been considered associated with periodontitis in recent years.[131] ^27 Furthermore, the presence of Anaeroglobus has been associated with symptomatic atherosclerosis, rheumatoid arthritis, and thyroid cancer.[132] ^30 , [133]^42 , [134]^43 Contrarily, it is generally accepted that Neisseria is part of the normal oral flora. The abundance of Neisseria showed decreasing trends from C1 through C2 to A in this study, suggesting that Neisseria may play a positive role in oral and cardiovascular health. Neisseria can facilitate the conversion of nitrate and nitrite into nitric oxide, a free radical possessing antimicrobial properties that can inhibit anaerobes implicated in periodontal diseases and involved in the maintenance of metabolic and cardiovascular homeostasis.[135] ^44 , [136]^45 Taken together, the results showed that periodontitis may be a potential risk factor for atherosclerosis, in which oral microbiota may play a significant role. Revealing the metabolomic characteristics of salivary microbiota is helpful to deeply understand how microbial changes affect the occurrence and development of CAS. The pathway enrichment analysis showed that glycerophospholipid and sphingolipid metabolism were altered in patients with CAS compared with controls. Glycerophospholipid metabolism is a key pathway closely related to cardiovascular disease.[137] ^46 Glycerophospholipids are the major structural lipids of membranes, of which common species are phosphatidyl choline, phosphatidyl serine, phosphatidyl ethanolamine, and phosphatidyl inositol. In recent years, studies have indicated that altered levels of specific glycerophospholipid metabolites are characteristic of cardiovascular risks.[138] ^47 In this study, an increase in lysophosphatidyl cholines (15:0/0:0) was detected in CAS, and they were positively correlated with several disease‐associated genera and IMT. Lysophosphatidyl cholines are primarily catalyzed by phospholipase A1 or A2 from phosphatidyl choline. Lysophosphatidyl cholines can induce the migration of lymphocytes and macrophages, enhance the generation of proinflammatory cytokines, induce oxidative stress, and facilitate apoptosis, thus promoting the progression of atherosclerosis and other cardiovascular disorders.[139] ^48 Sphingolipids are biologically active lipids with important roles in various cellular processes. In this study, we observed sphingolipid metabolism disorders and decreases of sphingosine, sphinganine, and phytosphingosine in patients with CAS, consistent with previous research.[140] ^49 , [141]^50 Sphingolipids produced by the important periodontal pathogen Porphyromonas gingivalis have been detected in carotid atherosclerotic lesions, which suggests that bacterial sphingolipids may play a potential role in cardiovascular disease.[142] ^51 In particular, the increase of TMAO in the saliva of patients with CAS is of interest, because this metabolite has been found strongly associated with atherosclerosis and other cardiovascular diseases. TMAO is a gut microbial‐host cometabolite of dietary choline and carnitine that plays important roles in atherosclerosis.[143] ^7 Specifically, TMAO induces monocyte adhesion, facilitates the upregulation of macrophage scavenger receptors, diminishes tissue repair, and upregulates the production of proinflammatory cytokines.[144] ^52 The source of TMAO in the saliva remains unknown. We speculate that oral microbiota may alter the composition and abundance of gut microbiota by mouth–gut axis and thus influence the metabolism in the circulation. Another upregulated metabolite, LTD4, is a lipid mediator that may promote inflammation. LTD4 was found to be positively correlated with IMT in this study. Furthermore, an increase of LTD4 was associated with increased abundance of Streptococcus, Tannerella, and Anaeroglobus. Scholars generally accept that oral bacteria can stimulate inflammatory cells to produce a large number of inflammatory cytokines and other mediators in periodontal disease and spill into the blood, thus promoting the formation and development of atherosclerosis. The positive correlations between LTD4 and CAS‐enriched genera further supported the role of microbiota‐mediated inflammation in disease pathogenesis. However, this study revealed a notable decrease in the saliva carnosine levels of patients with CAS, which exhibited a negative correlation with IMT. Carnosine is an endogenous dipeptide consisting of β‐alanine and l‐histidine, and it is also a dietary ingredient abundant in foods such as chicken and fish. Barski et al. found that carnosine can inhibit atherosclerosis by preventing the oxidation of low‐density lipoprotein‐cholesterol and facilitating the clearance of aldehydes generated by lipid peroxidation. Additionally, dietary carnosine supplements can prevent the formation of atherosclerosis lesions in apolipoprotein E^−/− mice.[145] ^53 Consequently, dietary carnosine supplementation may serve as an efficacious strategy for the prevention and management of atherosclerosis. Inflammation is widely recognized as a key factor in the progression of atherosclerosis and related cardiovascular diseases. Periodontitis‐associated systemic inflammation likely arises from hematogenous dissemination of periodontal bacteria or spillover of inflammatory mediators from periodontal tissues to the bloodstream.[146] ^54 To investigate the potential correlation between salivary microorganisms, metabolites, and systemic inflammation, the levels of systemic inflammatory markers in serum were determined. Hs‐CRP has garnered significant attention as a systemic inflammatory marker due to its application in screening and predicting cardiovascular disease risk. In this study, it was found that the serum levels of hs‐CRP were significantly higher in the group with CAS than those in control group, and the levels increased gradually in C1, C2, and A groups, indicating that severe periodontitis and CAS were closely associated with systemic inflammation. Correlation analysis revealed a significant relationship between serum hs‐CRP levels and the presence of Anaeroglobus and Neisseria, as well as the inflammatory mediator LTD4 in saliva. These findings suggest that systemic inflammation associated with severe periodontitis may be driven by oral microorganisms, thereby supporting an indirect inflammatory pathway linking periodontitis to cardiovascular disease. Research has indicated that cardiovascular disease remains a substantial burden in Liaoning Province, especially in rural areas. Specifically, the premature mortality rate of cardiovascular disease in urban areas of Liaoning Province was recorded at 10.44% in 2017, whereas it was even higher in rural areas, reaching 11.10%, and projections estimate a rise to 14.73% by 2030.[147] ^55 In this study, the study population comprised individuals residing in rural areas in northeast China who did not receive any therapeutic intervention, thereby contributing to a better understanding of the natural progression of the disease. Furthermore, the study participants were selected from a cohort of 1933 subjects who underwent carotid plaque screening and were matched based on cardiovascular disease risk factors, thus facilitating the elucidation of the association between oral microorganisms, metabolites, and CAS. Additionally, our study also revealed a high prevalence of hypertension, diabetes, dyslipidemia, overweight, and other traditional risk factors in the study participants. There are multiple potential factors contributing to this phenomenon. First, the cold climate and dietary habits characterized by high salt and fat intake of northeast China may play a role. Second, there is a noticeable lack of awareness, treatment, and control of related risk factors in rural areas. In comparison to urban areas, the lack of awareness regarding dental care and inadequate management of risk factors in rural areas of Northeast China may be closely related to the heavy burden of cardiovascular disease. Consequently, it is imperative to manage the risk factors associated with cardiovascular disease. In the present study, a significant correlation between severe periodontitis and CAS was observed, and periodontal pathogens may be involved in the pathogenesis of CAS. Consequently, we should pay more attention to the associations between periodontitis and atherosclerotic cardiovascular disease. It is recommended that patients with periodontitis receive adequate periodontal therapy, undergo cardiovascular disease risk assessment, and actively manage the associated risk factors. However, it is noteworthy that the control group exhibited a significant prevalence of severe periodontitis. The reason for severe periodontitis patients not afflicted with CAS may be associated with the low abundance of Streptococcus, Cutibacterium, Lactobacillus, and LTD4, coupled with the high abundance of Neisseria and carnosine. Therefore, risk screening through detecting of related microorganisms and dietary carnosine supplementation may hold considerable importance in the prevention and management of cardiovascular disease. However, we acknowledge the following limitations of this study. First, the use of traditional differential abundance analysis appears to exhibit false positive identification to some extent, and then we will further confirm these findings through clinical studies with large sample sizes.[148] ^56 Second, the results of stratification analyses according to age in the sensitivity analysis did not show a significant difference, potentially attributable to the influence of confounding variables, and may also indicate that the abundance of differential genera and metabolites varied with age. Third, the number of patients with CAS and gingivitis or mild to moderate periodontitis was too small, which prevents further persuasive subgroup analyses to elucidate the effect of periodontitis on CAS. Conclusions In conclusion, this study discovered a strong correlation between oral microorganisms and metabolites and the occurrence of CAS, and severe periodontitis may be a noteworthy risk factor for CAS. It is necessary to conduct multiomics studies with larger sample sizes from multiple centers to validate the findings of our research. Sources of Funding This work was supported by the National Key Research and Development Program of the Ministry of Science and Technology of China (No. 2023YFC2506302). Disclosures None. Supporting information Tables S1–S4 Figures S1–S7 [149]JAH3-13-e034014-s001.pdf^ (880.3KB, pdf) Acknowledgments