S. Thaddeus Connelly, DDS, MD, PhD, FACS
San Francisco Veteran's Affairs Healthcare System
University of California San Francisco • GengyeUSA • 2026
This presentation distinguishes established evidence from supported mechanisms and working hypotheses that still require causal testing.
- Periodontitis generates systemic inflammatory and microbial exposure
- Periodontal disease is associated with ASCVD and chronic outcomes
- Periodontal therapy can improve endothelial or inflammatory biomarkers
- P. gingivalis OMVs carry virulence cargo and injure host cells in models
- OMVs can promote endothelial activation and mitochondrial stress
- P. gingivalis is associated with certain cancers and neurodegenerative pathways
- A conserved OMV-driven program contributes across vascular, tumor, and brain disease
- Salivary/blood OMV signals can serve as scalable upstream biomarkers
- Upstream oral intervention can change long-term systemic trajectories
The opportunity is not to collapse diseases into one disease, but to identify shared upstream interfaces worth measuring and testing.
Cardiovascular disease is the leading cause of global mortality
Cancer accounts for nearly 1 in 6 deaths worldwide
Dementia is a major cause of disability and death globally
Even a modest upstream risk signal could matter clinically if it is measurable and modifiable.
Periodontitis creates a persistently inflamed, ulcerated pocket surface — providing repeated opportunities for bacterial products, inflammatory mediators, and extracellular vesicles to enter the circulation.
- Smaller and more mobile than whole bacteria
- Package concentrated virulence cargo
- Can enter host cells and alter signaling without viable bacterial colonization at the distal site
- Provide a tractable biomarker and intervention interface
20–250 nm bilayer vesicles shed from Gram-negative bacteria. They package concentrated virulence factors and move through tissues and fluids more easily than intact bacteria.
- Gingipains
- Heterogeneous LPS / lipid A species
- PPAD and adhesins / fimbrial components
- Nucleic acids
Cargo composition matters — different components likely drive different phenotypes.
Specifying which cargo class drives which phenotype — and where evidence is strongest — tightens the biology.
Useful for organizing experiments; risky if presented as already-proven causality.
Barrier injury is the gatekeeping step that makes distal effects plausible. If OMVs cannot alter barriers, the distal-disease narrative gets much weaker.
P. gingivalis cells and OMVs increase CXCL8 and E-selectin signaling, promote monocyte adhesion, and perturb endothelial integrity and inflammatory tone.
Long-term P. gingivalis infection in primary oral epithelial cells induces EMT-like changes: E-cadherin loss and increases in vimentin, Slug, Snail, and ZEB1.
Oral microbial or host EVs can reach the brain in animal models and perturb BBB-related tight-junction signals. Human relevance is plausible but still incomplete.
A strong mechanistic node and reasonable translational target — but not yet proof of cross-disease efficacy.
In endothelial models, P. gingivalis exposure increases mitochondrial fragmentation and mtROS, lowers membrane potential, and decreases ATP via Drp1-dependent fission — a concrete bioenergetic mechanism beyond vague "inflammation."
- Vascular: Bioenergetic stress sensitizes endothelium to dysfunction, inflammatory signaling, and remodeling
- Neural: Pg OMVs can induce neuroinflammation, BBB tight-junction loss, tau phosphorylation, and memory effects in mice
Pg OMVs can induce ERK1/2 → Runx2 signaling, alkaline phosphatase activity, and osteogenic calcification in vascular smooth-muscle cells in vitro.
Chronic P. gingivalis exposure drives EMT-like programs linked to invasion and altered apoptosis — most convincing in oral and upper-GI contexts, not as a universal solid-tumor rule.
OMV exposure can amplify inflammatory phenotypes including NLRP3-related signaling and tau-associated changes in mouse models. Human translation remains uncertain.
- P. gingivalis infection can alter host EV release and cargo
- EVs can propagate tau-related pathology in AD models
- Mitochondrial transfer occurs in cancer and can affect chemoresistance, immune biology, and metabolic fitness
- Whether OMV injury triggers a conserved EV/organelle relay across vascular, tumor, and neural tissues
- Which transfer events are harmful versus reparative
- How to block pathological transfer without suppressing beneficial repair
- Whether these pathways are measurable and clinically actionable in humans
OMV-induced VSMC calcification is real in vitro and clinically relevant. Microcalcification may destabilize lesions; macrocalcification can reflect healing or stabilization — nuance matters.
Best P. gingivalis evidence is in oral and selected GI/upper-GI cancers. Enough biology to take seriously — not enough to claim a universal solid-tumor mechanism.
OMV exposure can worsen neuroinflammation and tau-related changes in animals. Direct human causal proof remains absent. Describe as association plus biological plausibility.
Be ambitious in screening and collaboration; be conservative in causal claims and treatment promises.
- Treat periodontitis as a systemic health issue, not just a local dental problem
- Capture periodontal status in cardiometabolic and geriatric histories where relevant
- Expect local benefit and some biomarker or endothelial improvement after periodontal therapy
- Claiming periodontal disease is a fully validated causal driver of ASCVD, cancer, or AD
- Telling patients gingipain inhibition or oral therapy is proven to slow dementia
- Presenting the AHA statement as if it resolved the causal question
Target engagement and subgroup signals make the biology interesting. However, GAIN missed its co-primary endpoints in the overall cohort.
Atuzaginstat is a tractability signal — not clinical validation of the OVN thesis.
Build integrated dental-medical workflows, biomarker panels, and prospective studies that measure whether treating the mouth moves systemic biology.
That is a stronger story than claiming one hidden cause of three diseases.
The most defensible strategy: reduce source burden and prove biomarker movement first.
Highest-credibility path: periodontal therapy, biofilm disruption, precise subgingival delivery, and oral-care formulations that measurably reduce P. gingivalis burden or virulence output.
Gingipain inhibition is the clearest cargo-target concept. Vaccine, antibody, and OMV-binding decoy ideas are attractive but early — platform possibilities, not established therapies.
NLRP3 and mitochondrial-resilience approaches may blunt downstream injury. Position as adjunctive pathway ideas unless paired with source-control biomarkers.
Useful pathway-adjacent bets — but not proof of the OVN thesis.
SNF472 slowed coronary artery calcium progression in hemodialysis phase 2b. Its phase 3 calciphylaxis program did not meet primary endpoints — pathway-adjacent evidence, not a validated cardiovascular OVN therapy.
Exosome inhibition, EV decoys, and nanoparticle "sponges" are conceptually attractive. They remain early, and their value is strongest when paired with a measurable pathogenic EV signal.
Tunneling-nanotube or gap-junction blockade may matter in oncology, but selectivity is the central challenge — some mitochondrial transfer is reparative rather than pathological.
Goal: move from plausible narrative to measurable, targetable biology.
Quantify OMV load, gingipain activity, antibodies, and inflammatory markers; correlate with hsCRP, endothelial markers, CAC/IMT, and cognition.
Compare wild-type vs. gingipain-deficient vs. PPAD-deficient OMVs across endothelium, VSMCs, microglia, and tumor cells.
Run the same OMV prep through a BBB chip, vascular coculture, and tumor-invasion model to test whether the cascade is actually conserved.
Measure salivary and systemic changes before and after periodontal therapy, with and without an adjunct source-control formulation.
Test whether baseline oral biomarker signatures predict vascular progression, frailty, or cognitive decline better than dental measures alone.
The smartest early businesses do not require proving full cross-disease causality first.
RUO → CLIA/LDT path; recurring consumables and a biomarker data asset.
Combine periodontal status, biomarkers, and referrals into workflow software for dental and medical groups.
Office-dispensed rinses, gels, or adjuncts positioned around source reduction plus biomarker tracking.
Identify oral-pathogen-high patients for neurology, inflammation, or oncology trials.
Turn real-world dental-medical data into publications, payer leverage, and future defensibility.
Oral bacterial EVs are plausible systemic effectors and a tractable upstream biomarker interface — strong enough to justify measurement, intervention studies, and disciplined translational development now.
Periodontitis ↔ ASCVD association; OMV endothelial effects; mitochondrial stress; selected neural and cancer-related model systems.
Human causality, cargo attribution, prospective biomarker validation, and proof that the proposed cascade is truly conserved across tissues.
Winning Strategy: Prove an upstream signal, show that intervention moves it, and only then claim disease modification. That path is scientifically stronger and commercially more credible.
Source for the ~32% global-mortality figure used to frame vascular burden.
Source for the "nearly 1 in 6 deaths" cancer-burden statement.
Supports describing dementia as a major cause of disability and death.
Establishes periodontal disease as a globally important health and wellbeing burden.
AHA statement anchoring the deck's association-focused ASCVD framing.
Umbrella review supporting a real but still non-causal periodontitis-CVD signal.
These sources support the strongest mechanistic claims in the endothelial / barrier section.
Shows PPAD activity can promote OMV biogenesis and surface translocation.
Key cautionary paper: P. gingivalis lipid A interacts variably with TLR2 and TLR4.
Direct evidence for endothelial activation induced by P. gingivalis and derived OMVs.
Important barrier paper: OMVs increase vascular permeability.
Supports Drp1-dependent mitochondrial fission, mtROS, and ATP loss in endothelial models.
Most on-point paper for VSMC osteogenic differentiation and calcification via ERK1/2-Runx2.
This group supports the deck's more cautious framing around BBB biology, AD-like effects, EMT, and selected cancer contexts.
Shows long-term P. gingivalis infection can drive EMT-like epithelial changes.
Model-system evidence that orally induced small EVs can cross the blood-brain barrier.
Key mouse-model paper for NLRP3 activation, tau phosphorylation, and memory effects after OMV exposure.
Helps narrow the cancer claim to oral / orodigestive squamous-cell settings.
Important guardrail: AD relevance is biologically plausible, but direct human causality is unproven.
Frames mitochondrial transfer as a real oncology mechanism and possible target.
These sources explain why the revised deck separates tractable biology from proven clinical efficacy.
Seminal human study showing periodontal treatment can improve endothelial-function-related readouts.
Supports describing dementia links as association plus biological plausibility rather than settled causality.
Useful reference for gingipain inhibition as a tractable target concept.
Phase 2b evidence that SNF472 slowed coronary calcium progression in hemodialysis.
Phase 3 CALCIPHYX result — keeps the deck honest about pathway-adjacent rather than validated therapy.
Important counterweight: mitochondrial transfer can be beneficial, so blocking it requires selectivity.
Bacterial extracellular vesicles as a candidate oral–systemic disease interface