Deciphering Periodontitis: Host Mechanisms, Predictive Biomarkers, and Systemic Drivers for Precision Prevention

Periodontitis afflicts over 3 billion individuals worldwide. Its resistance to treatment and high recurrence rate remain unresolved by current paradigms, placing immense burdens on elderly health, policy making, and healthcare expenditure. Here, we present the most comprehensive host‑centric analysis of chronic periodontitis to date. This integrates 115,000 gingival single-cell transcriptomes with nationally representative NHANES epidemiological data via an AI-powered machine learning pipeline. Our single-cell atlas reveals four key alterations in chronic periodontitis: a substantial reduction in epithelial barrier integrity, a significant redistribution toward memory-desert T cells, a pronounced expansion of plasmacytes, and predominant infiltration of M1 macrophages. These alterations are all driven by a matrix-centered intercellular network that amplifies extracellular matrix disruption, chemotactic signaling, and angiogenesis. These findings highlight stress-response and barrier-inflammation pathways across pseudotime. In parallel, a 20-gene machine-learning signature distinguishes PD tissue from healthy gingiva and supports risk modeling at the population level. Causal inference implicates uncontrolled systemic disorders and micronutrient deficiencies as key modifiable drivers. This multilayered framework supports the reconceptualization of chronic periodontitis as an inflammatory indicator of systemic health. It also informs future strategies for precision diagnostics and targeted therapies.