The research protocol was approved by the Committee for the Protection of Human Subjects of the University of Texas Health Science Center at Houston (UTHealth at Houston) (IRB number: HSC-DB-17-0636). Candidates were screened during routine dental visits at the clinic of School of Dentistry at UTHealth at Houston between 2018 and 2021. Individuals aged 21–75 with self-reported ethnicity/race of non-Hispanic African American (AA), non-Hispanic Caucasian American (CA), or Hispanic American (HA) were enrolled after initial periodontal examination. The examination documented plaque index (PI), bleeding on probing (BOP), probing depth (PD), clinical attachment level (CAL), furcation involvement, tooth mobility, and keratinized tissue on all teeth (Newman et al., 2018). Radiographs were taken during this screening phase to assess bone loss. The clinical periodontal examinations were performed by trained dental examiners who are faculty members of the School of Dentistry, UTHealth at Houston. All study participants were diagnosed with generalized periodontitis Stage II or III, regardless of their grading, based on the 2017 World Workshop classification (Tonetti et al., 2018; Papapanou et al., 2018a). The enrolled patients also met the following criteria: ≤4 tooth loss due to periodontitis, interdental CAL ≥3mm and PD ≥5 mm at two or more teeth in different quadrants, and radiographic bone loss ≥15%. Other criteria for study participation were 1) no scaling and root planning within the previous year or periodontal surgeries in the previous five years; 2) no antibiotic therapy in the previous six months; 3) not pregnant. The periodontal parameters and diagnosis of participants were abstracted from the Electronic Health Record (axiUm). In addition, medical history and dental history were extracted from axiUm and recorded by a dental student (LA) who is blinded to study design and bacterial data. Cardiovascular diseases (including infective endocarditis, coronary artery disease, heart attach, congestive heart failure, hypertension, cardiac arrhythmia, rheumatic fever, stroke, mitral valve prolapse), respiratory diseases (chronic bronchitis or emphysema, asthma), sleep apnea, renal failure, gastrointestinal diseases (including hepatitis, stomach ulcer, crohn’s disease, colitis, liver disease), endocrine diseases (diabetes, thyroid disorder), hemotologic disorders, arthritis, osteoporosis, depression, autoimmune diseases, cancer, usage of alcohol or tobacco, dry mouth, and habitually clench or grind teeth were recorded as present or abscent. Body mass index (BMI) was recorded as numerical numbers. Neiberhood median incomes of the participants were decided based on their residential zip codes (https://www.incomebyzipcode.com/).

Dental plaque samples were collected by board-certified periodontists using sterile paper points at baseline prior to any dental treatment. The samples were labelled with numbers according to sampling sequence. The paper points were placed in ≥5mm pockets in different quadrant for 1 minute and then immersed immediately in an Eppendorf tube with 0.5 ml of Tris-EDTA (TE) buffer (pH 7.5) (Wang et al., 2009). Oral bacteria were harvested by centrifugation and the bacterial pellets were resuspended in 100 µl TE buffer. Chromosomal DNA was released by two cycles of freezing at -80^°C overnight and boiling for 10 minutes.

All bacteria tested were enumerated by qPCR using SYBR Green PCR mix (Bio-Red Laboratories Inc., Redmond, WA, USA) with species-specific primers listed in Table 1 . P. gingivalis strains with different fimA genotypes were identified by qPCR using strain-specific primers (Zheng et al., 2011). Levels of total bacteria were determined using primers corresponding to the conserved sequences of cyanobacterial small subunit rRNA genes (Turner et al., 1999) ( Table 1 ). Standard curves used to enumerate bacterial cells were generated by qPCR using the genomic DNA from each bacterial species as previously described (Wang et al., 2009). Bacterial ratios were calculated using the numerical levels of two bacteria from the same sample. Bacterial quantitation by qPCR was performed by a designated technician blinded to subject information.

Continuous variables were analyzed using one-way ANOVA. The chi-squared test or Fisher’s exact test was performed for categorical variables. Levels of periodontitis-associated bacteria were categorized into two groups by their medians and compared among the three study groups. Stratified analyses were conducted to evaluate the differences in levels of periodontitis-associated bacteria between perodontitis stages II and III. Correlations between levels of bacterial species (r) were determined using Pearson correlation coefficient. P-values <0.05 were considered statistically significant. SAS version 9.4 (SAS Institute, Cary, NC, USA) was used to conduct all statistical analyses.

We enrolled 191 periodontitis patients in this study, including 56 AAs, 67 CAs, and 68 HAs with the mean age of 51.66 ± 12.79 years. Diagnosis creteria for the patients were based on the 2017 World Workshop classification (Tonetti et al., 2018; Papapanou et al., 2018b). Among them, 61.78% were diagnosed with periodontitis stage III and 38.22% were diagnosed with stage II. There was no significant difference in the distribution of periodontitis stages among AAs (66.04% with stage III), CAs (54.41% with stage III), and HAs (67.64% with stage III) (p = 0.291). In addition, there was no significant difference among the racial and ethnic groups with regards to gender (p = 0.228). To investigate the possible existence of social inequalities among these racial/ethnic groups, we used the participants’ residential zip codes to obtain median incomes of the neighborhoods in which they reside. Our results showed the neighborhood median income of the CA group to be significantly higher ($73,895 ± $26,906) than that of the AA or HA group ($58,013 ± 22,470 and $58,558 ± $ 21,434, respectively) (p<0.001). Our observation agrees with a previously reported likelihood of an association between neighborhood socioeconomic circumstances and periodontitis (Borrell et al., 2006). However, further studies are needed to elucidate how economic conditions of a neighborhood impact periodontal health of its residents.

We also found significant differences in levels of BOP among the racial/ethnic groups. We observed the highest degree of BOP in the HA group (44.18% ± 25.80%), followed by the AA group (41.91% ± 25.74%) and the CA group (38.39% ± 24.51%) (p = 0.007) ( Table 2 ). However, we did not observe significant differences in the levels of PI among AAs (66.61% ± 32.27%), CAs (61.65% ± 24.57%), and HAs (64.77% ± 27.57%) (p = 0.569). These results suggest that different profiles of microbiota likely led to a more severe inflammatory response found in AA and HA patients compared to CA patients.

To investigate the oral microbial profiles of the study subjects, we used qPCR to measure the distribution and levels of several well-studied oral bacteria in their dental plaque samples, including keystone pathogens, accessory pathogens, and pathobionts (Hajishengallis and Lamont, 2016). Since the data are not evenly distributed, we designated bacterial levels as either higher or lower than their medians. As shown in Table 3 , total bacterial levels in the dental plaque samples were determined using primers corresponding to cyanobacterial 16S rRNA as probes (Turner et al., 1999). Approximately 50% of all samples had more than 10⁹ total bacterial cells. However, 71% of samples from the AA patients had more than 10⁹ total bacterial cells, which was much higher than that for the CA patients (47%) and for the HA patients (38%) (p = 0.0009). Using qPCR, we detected P. gingivalis in all 191 samples. Higher levels of P. gingivalis (>10⁶) was detected in 54.41% of samples from the HA patients and in 48.21% of samples from the AA patients, but in only 24.88% of samples from the CA patients (p = 0.0008), which may be linked to higher BOP index observed in the HA and AA patients. We also detected Tannerella forsythia and Fusobacterium nucleatum in all samples, while Treponema denticola was present in 66.49% of the samples. These three species were all evenly distributed among the AA, CA, and HA groups.

Our previous study on 13 subjects with periodontitis demonstrated that there appeared to be an inverse relationship between the number of S. cristatus versus P. gingivalis cells in dental plaque (Wang et al., 2009). Therefore, we postulated that higher levels of S. cristatus, an early colonizer of oral microbial communities, may control and suppress P. gingivalis levels, and that the higher risk of periodontitis in AA and HA populations is, at least in part, dependent on a different microbial composition containing less S. cristatus. Here, we found that 50.75% of dental plaque samples from the CA patients had S. cristatus/P. gingivalis ratios greater than 100, compared to 41.07% of samples from the AA patients and 27.94% of those from the HA patients (p = 0.025) ( Table 3 ). Further analysis of the S. cristatus/P. gingivalis ratios revealed a significant difference in the levels of P. gingivalis and T. denticola between dental plaque samples with higher S. cristatus/P. gingivalis ratios and samples with lower ratios ( Table 4 ). Particularly, levels of P. gingivalis were approximately 700 time higher in samples with relatively lower levels of S. cristatus. We also observed a slight increase in the abundance of total bacteria, T. denticola, and F. nucleatum in samples with lower S. cristatus/P. gingivalis ratios. These results further support our postulation that S. cristatus benefits the host by antagonizing the colonization and accumulation of P. gingivalis (Wang et al., 2009).

In addition, we examined the distribution of P. gingivalis of different fimA genotypes (types I and Ib–V), which are classified based on their nucleotide sequences of the fimA gene. We found P. gingivalis of fimA genotype II to be present in 55.5% of all samples. This is consistent with previous reports stating P. gingivalis type II as the predominant P. gingivalis strain found in periodontitis patients (Amano et al., 1999; Enersen et al., 2008). We also detected the type IV strain in 20.94% of all samples, followed by type III (10.47%), type I (9.94%), type Ib (2.10%), and type V (2.09%) ( Table 5 ). We did not observe any significant difference in the distribution of the different fimA strains among the three racial/ethnic groups and between patients with periodontitis stage II and III.

To examine the relationship between P. gingivalis levels and periodontitis progression as well as the abundance of other bacteria, we sequentially analyzed the keystone and accessory pathogen levels in patients with different periodontitis stages and characteristics. As shown in Table 6 , more samples from patients with periodontitis stage III exhibited higher levels of total bacteria (54.17% >10⁹) and P. gingivalis (44.17% >10⁶) than samples from stage II patients, the latter with 46.48% >10⁹ for total bacteria and 38.03% >10⁶ for P. gingivalis. However, these differences are not statistically significant. We also observed a similar trend for the S. cristatus/P. gingivalis ratios. Specifically, 46.48% of samples from stage II patients had ratios greater than 100, but only 35.83% of samples from stage III patients had ratios greater than 100. We observed significantly higher levels of T. denticola in samples from stage III patients than in those from stage II patients (66.28% vs. 43.90% >10³, p = 0.016). These results are consistent with our observation that samples with higher T. denticola levels also possessed higher levels of P. gingivalis (mean = 8.86 × 10⁷) than samples with lower T. denticola levels (mean = 5.10 × 10⁶) (p = 0.0078) ( Table 4 ). These findings indicate that co-colonization of P. gingivalis and T. denticola may be associated with periodontitis progression. We also examined correlation between other independent variables, including body mass index, smoking, and diabetes, and severities of periodontitis. We did not found significant influence of these variables on stages of periodontitis in this cohort ( Table 6 ).

In addition, we examined the correlations among these oral bacteria. Levels of P. gingivalis, T forsythia, T. denticola, and F. nucleatum in the samples positively correlated with the level of total bacteria to various extents ( Table 7 ). The level of P. gingivalis correlated most strongly with the level of total bacteria (r = 0.66), followed by T. forsythia (r = 0.59), F. nucleatum (r = 0.29), and T. denticola (r = 0.18). Interestingly, T. forsythia exhibited a relatively high degree of correlation with P. gingivalis, T. denticola, and F. nucleatum (r > 0.45), suggesting the importance of T. forsythia in periodontal dysbiosis. In contrast, P. gingivalis showed weaker correlations with T. denticola (r = 0.18) and F. nucleatum (r = 0.15). We did not observe any significant correlation between T. denticola and F. nucleatum.