Background/Objective: Periodontitis is a chronic inflammatory disease with a multifactorial etiology involving clinical, behavioral, and microbial determinants. However, the relative contribution of these factors to disease discrimination remains debated, particularly in untreated adult populations. This study is aimed to characterize the clinical, epidemiological, and microbial features associated with periodontitis in an adult cohort, and to compare the discriminatory performance of microbiome-based predictive models with conventional clini-cal-behavioral models. Methods: A cross-sectional study was conducted in 943 adult participants. Periodontal status was determined by experienced clinicians according to the diagnostic criteria established in the 2017 World Workshop on the Classification of Periodontal and Peri-implant Diseases and Conditions, based on clinical attachment loss (CAL), probing depth (PD), bleeding on probing (BOP), and radiographic bone loss. Clinical variables, behavioral factors (smoking, bruxism, diet), intraoral conditions (caries and malocclusion), and systemic comorbidities were recorded. The salivary microbiome was analyzed using targeted multiplex PCR for selected periodontal bacterial species. Predictive models were constructed using logistic regression and least absolute shrinkage and selection operator (LASSO) variable selection. Results: Periodontitis was diagnosed in 47.2% of participants. Age, smoking, and bruxism were significantly associated with periodontitis, although their combined discriminatory capacity was modest (AUC = 0.65). Malocclusion emerged as the only significant intraoral predictor (OR = 2.00). Microbiologically, individuals with periodontitis exhibited increased levels of recognized periodontopatho-gens, including P. gingivalis, T. forsythia, and E. corrodens, along with reduced levels of commensal species such as S. mutans. Microbiome-based models demonstrated superior discriminatory performance (AUC = 0.76, LASSO). E. corrodens and C. sputigena were independently associated with greater probing depth (p < 0.001). Additionally, E. corrodens levels were significantly higher in periodontitis patients with cardiovascular disease compared to controls. Conclusions: While conventional clinical and behavioral variables show limited capacity to discriminate periodontitis status, microbiome-based predictive models—particularly at the species level—demonstrate improved diagnostic performance. These findings support the potential role of salivary microbial signatures as adjunctive, non-invasive biomarkers reflecting periodontal inflammatory status.

Background/Objective: Periodontitis is a chronic inflammatory disease influenced by clinical, behavioral, and microbial determinants. However, the contribution of these factors to disease remains a topic of debate, particularly in untreated adult populations. This study aims to characterize the clinical, epidemiological, and microbial features associated with periodontitis in an adult cohort and to compare the discriminatory performance of microbiome-based predictive models with conventional clinical–behavioral models. Methods: A cross-sectional study was conducted in 943 adults. Periodontal status was determined by experienced clinicians according to the 2017 World Workshop on the Classification of Periodontal and Peri-implant Diseases and Conditions. Clinical variables, behavioral factors (smoking, bruxism, diet), intraoral conditions (caries and malocclusion), and systemic comorbidities were recorded. The oral microbiome was analyzed using targeted PCR for selected periodontal bacteria. Predictive models were constructed using logistic regression and least absolute shrinkage and selection operator (LASSO) variable selection. Results: Periodontitis was diagnosed in 47.2% of participants. Age, smoking, and bruxism were significantly associated with periodontitis. Malocclusion was the only significant intraoral predictor (OR = 2.00). Individuals with periodontitis exhibited increased levels of periodontopathogens, including P. gingivalis, T. forsythia, and E. corrodens, along with reduced levels of S. mutans. Microbiome-based models demonstrated superior discriminatory performance (AUC = 0.76, LASSO). E. corrodens and C. sputigena were independently associated with greater probing depth (p < 0.001). Conclusions: Microbiome-based predictive models, particularly at the species level, showed better discriminatory performance than conventional clinical–behavioral models. These findings support the potential utility of salivary microbial signatures as adjunctive, non-invasive biomarkers of periodontal inflammatory status.