Inflammation of the periodontium associates with risk of future cardiovascular events

Abstract

Background:

While growing evidence suggests a link between periodontal disease (PD) and cardiovascular disease (CVD), the independence of this association and the pathway remain unclear. Herein, we tested the hypotheses that: (1) inflammation of the periodontium (PD_inflammation) predicts future CVD independently of disease risk factors shared between CVD and PD, and (2) the mechanism linking the two diseases involves heightened arterial inflammation.

Methods:

¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG-PET/CT) imaging was performed in 304 individuals (median age 54 years; 42.4% male) largely for cancer screening; individuals without active cancer were included. PD_inflammation and arterial inflammation were quantified using validated ¹⁸F-FDG-PET/CT methods. Additionally, we evaluated the relationship between PD_inflammation and subsequent major adverse cardiovascular events (MACE) using Cox models and log-rank tests.

Results:

Thirteen individuals developed MACE during follow-up (median 4.1 years). PD_inflammation associated with arterial inflammation, remaining significant after adjusting for PD and CVD risk factors (standardized β [95% CI]: 0.30 [0.20–0.40], P < 0.001). PD_inflammation predicted subsequent MACE (standardized HR [95% CI]: 2.25 [1.47 to 3.44], P <0.001, remaining significant in multivariable models), while periodontal bone loss did not. Furthermore, mediation analysis suggested that arterial inflammation accounts for 80% of the relationship between PD_inflammation and MACE (standardized log odds ratio [95% CI]: 0.438 [0.019–0.880], P = 0.022).

Conclusion:

PD_inflammation is independently associated with MACE via a mechanism that may involve increased arterial inflammation. These findings provide important support for an independent relationship between PD_inflammation and CVD.

1 |. INTRODUCTION

Periodontal disease, an inflammatory condition impacting the supporting structures of teeth,¹ afflicts 42% of dentate US adults aged ≥30 years, with 8% having severe periodontitis. Overall, 3% of all periodontally probed sites (9% of all teeth) had periodontal probing depth (PD) of ≥4 mm, and 19% of sites (37% of teeth) had clinical attachment loss of ≥3 mm. Severe periodontitis was most prevalent among adults aged ≥65 years, Mexican Americans, non-Hispanic Blacks, and smokers. A large body of epidemiological evidence links periodontal disease to cardiovascular disease (CVD).^2–5 However, it remains unclear whether that relationship is causal, whereby the chronic inflammatory milieu elicited by periodontitis potentiates inflammatory conditions like CVD,^6,7 or is non-causal, whereby periodontitis and CVD simply share risk factors like smoking, diabetes mellitus, age, sex, and genetic background.^4,8,9

Standard quantitative assessment of periodontitis includes measures of attachment loss and pocket depth, which provide an index of accumulated damage to the periodontium. However, these traditional structural/anatomical measures of periodontitis might not accurately reflect the ongoing biological processes that drive periodontal disease progression. Alternatively, PD_inflammation, which initially prompts bone loss and pocket formation,¹⁰ may be a more appropriate measure of ongoing periodontal disease activity. Indeed, inflammatory biomarkers in gingival crevicular fluid correlate well with periodontal disease progression (disease activity).^11,12 Periodontal inflammation has repeatedly been shown to associate with arterial inflammation,^13,14 which may be particularly important, since arterial inflammation associates with major adverse cardiovascular events (MACE) risk.¹⁵ However, it remains untested whether PD_inflammation associates with MACE.

¹⁸F-fluorodeoxyglucose positron emission tomography/computed tomography (¹⁸F-FDG-PET/CT) is used clinically to identify metabolically active tissues non-invasively and is useful for the identification of inflamed or infected tissues as well as for measuring atherosclerotic inflammation.^16–18 The uptake of ¹⁸F-FDG-PET/CT is a reliable marker for the measurement of tissue glycolysis and ¹⁸F-FDG-PET/CT imaging has furthermore been validated as an effective means of measuring active periodontal inflammation^14,18–21 and evaluating anti-inflammatory therapies.^14,22 Accordingly, using ¹⁸F-FDG-PET/CT as the measure of PD_inflammation, we sought to: 1) test whether PD_inflammation independently associates with the risk for subsequent MACE and 2) evaluate whether arterial inflammation may mediate the link between periodontitis and CVD.

2 |. MATERIALS AND METHODS

2.1 |. Study population

Participants (N = 304) were selected from all individuals who underwent clinical ¹⁸F-FDG-PET/CT imaging, largely for cancer screening, at the Massachusetts General Hospital from 2005 through 2008. All individuals with adequate periodontal tissue imaging who were included in a prior study evaluating the relationship between arterial inflammation and MACE²³ were included in the current study (Fig. 1). Pre-defined inclusion criteria for study participants were: 1) absence of prior cancer or remission from cancer for at least 1 year before imaging and throughout the follow-up period, 2) absence of clinical CVD at the time of imaging, 3) absence of documented acute or chronic inflammatory or autoimmune disease or use of chronic anti-inflammatory treatment at the time of imaging or during the follow-up period, and 4) age >30 years. Those included had at least three clinical visit notes over at least 1 year for the assessment of baseline and follow-up status. The study protocol was approved by the Partners Human Research Committee.

FIGURE 1.

Flowchart of study participant selection. CVD, cardiovascular disease; CT, computed tomography; FDG, fluorodeoxyglucose; MGH, Massachusetts General Hospital; PET, positron emission tomography

2.2 |. ¹⁸F-FDG-PET/CT protocol

¹⁸F-FDG-PET/CT imaging was performed on a Biograph 64 (Siemens, Erlangen, Germany) or a similar system using previously validated approaches.^{22 18}F-FDG at a dose of 370 MBq was injected intravenously, and PET images were acquired in three-dimensional mode ≈60 minutes after ¹⁸F-FDG administration. Low-dose, non-enhanced CT (120 KVp, ~50 mAs) was performed for attenuation correction before the PET scan. The images were co-registered at a workstation (Leonardo–TrueD, Siemens Medical Solutions, Forchheim, Germany)²⁴ to facilitate image analysis.

2.3. ¹⁸F-FDG uptake measurement in the periodontal tissue

Images were blinded to patient identifiers and were analyzed using Leonardo TrueD software (Siemens, Forchheim, Germany). PD_inflammation was measured as the periodontal accumulation of¹⁸F-FDG, according to previously published methods.^13,14 The identification of the periodontal tissues was aided by use of the CT images; ¹⁸F-FDG uptake was evaluated on the co-registered images by manually drawing a rectangular region of interest (ROI) around the teeth from the mesial surface of the first premolar to the distal surface of the second molar carefully avoiding any spill-over from the surrounding tissues (e.g., pharyngeal and buccal mucosa and tongue). Four ROIs were drawn for each participant for the right and left mandibular and maxillary periodontal tissue, and the maximum standardized uptake values (SUVs) of ¹⁸F-FDG were registered for each quadrant. The SUV values for the periodontal quadrants were averaged to yield an average of the maximum periodontal SUVs.

2.4. ¹⁸F-FDG uptake measurement in the arterial wall

Arterial inflammation was assessed as a ratio of aortic wall to background venous blood metabolic activity. Circular ROIs were placed in the axial plane around the vessel wall every 3 mm starting 1 cm above the aortic valve annulus and continuing to the bottom of the aortic arch. Maximum SUV values were registered for each segment, and the mean of the SUV values was calculated for the entire ascending aorta. The background SUV was obtained by placing regions of interest in the superior vena cava in an area devoid of significant spillover activity from the surrounding tissues. Thereafter, a target-to-background-ratio was calculated by dividing the aortic SUV by the background venous SUV (Fig. 2).

FIGURE 2.

Periodontal and arterial ¹⁸F-FDG uptake. A) shows axial images at the level of periodontium from CT and ¹⁸F-FDG-PET-CT. ¹ B) shows coronal images of the ascending aorta. C and D) demonstrate box plots of periodontal inflammation and of atherosclerotic inflammation among individuals with versus without subsequent MACE. CT, computed tomography; ¹⁸F-FDG, ¹⁸F-fluorodeoxyglucose; PET, positron emission tomography

2.5 |. Periodontal bone loss assessment

Periodontal bone loss was measured using the attenuation correction CT images acquired before the PET scan. Evaluation of the periodontal CT images was performed by an experienced periodontist masked to all PET imaging and clinical data using previously validated methods.¹⁴ Alveolar bone resorption was assessed semi-quantitatively in each of the four dental quadrants and was graded as follows: edentulous, none, mild (limited to coronal one-third of the root), mild-to moderate, moderate (including the middle one-third of the root), moderate-to-severe, or severe (including the apical one-third of the root) bone loss, respectively. Classification of each individual’s disease severity was based on each individual’s most severely diseased quadrant. Periodontium of anterior teeth (incisors and canines) and third molars were not evaluated for bone resorption scoring.

2.6 |. Coronary calcium assessment

Coronary artery calcium (CAC) score was quantitatively assessed by a separate investigator who was blinded to clinical and PET data, using standardized methods.²⁵ A Hounsfield Unit (HU) threshold was pre-specified for this analysis: range of > 130 HU in more than three continuous voxels on non-gated CT images obtained from a hybrid PET/CT scanner. Since it is known that non-gated CT scans can underestimate the presence of CAC.^26–28 Accordingly, the lowest CAC scores (0 to 10) were combined into one group,²⁹ such that CAC was categorized into three groups: 0 to 10,11 to 99, and >100.

2.7 |. MACE adjudication

Two cardiologists masked to imaging data adjudicated MACE using the available clinical records. MACE was defined as a composite of CVD death, myocardial infarction, unstable angina, cerebrovascular accident, heart failure, or coronary or peripheral artery revascularization.³⁰

2.8 |. Statistical analysis

All analyses were performed using SPSS (IBM, Version 25.0, Armonk, NY). Data are presented as mean ± SD for continuous parametric variables, median (P25-P75) for continuous non-parametric variables, and frequency (%) for categorical variables. Pearson correlation coefficients were used to assess associations for normally distributed variables, and Spearman correlations were used for variables that are not normally distributed. Kaplan–Meier curves of MACE-free survival were generated, and log-rank tests were performed to determine statistical significance between groups. For these analyses, two approaches were employed to categorize PD_inflammation values as “higher” versus “lower”. The first approach identified “higher” values based on data distribution, as those PD_inflammation SUV values that are > 1 SD above the mean. The second approach used an ROC approach to define higher values as those that are more predictive of MACE, using the Youden method. The Youden index to identify the PD_inflammation value that best predicts MACE, based on the ROC analysis. This index gives equal weight to sensitivity and specificity and identifies the PD_inflammation SUV threshold that provides the greatest sum of sensitivity + (1-specificity). Univariable and multivariable Cox proportional hazards regression analysis were used (employing a continuous variable for PD_inflammation) to calculate hazard ratios (HRs) and 95% confidence intervals (CIs). Continuous variables were standardized. When >1 covariable was entered in the model, a backward stepwise (conditional) method was used.

Mediation (path) analysis was performed using the SPSS PROCESS macro, as previously described.³⁰ We examined the pre-specified single mediator path: (1) PD_inflammation ⟶ arterial inflammation ⟶ MACE. A two-sided P value <0.05 was considered statistically significant for all analyses.

3 |. RESULTS

3.1 |. Study participant characteristics

A total of 304 individuals were included with a median [P25 to P75] follow-up of 4.0 [3.0 to 5.1] years. The median age was 54.0 [44.0 to 63.8] years and 129 (41.6%) were males. Additional characteristics are listed in Table 1 and in Tables S1A and S1B in the online Journal of Periodontology.

TABLE 1.

Study participant characteristics

Characteristics	Full cohort (N = 304)
Age (years)	54.0 (44.0 to 63.8)
Male (%)	129 (42.4%)
Body mass index (kg/m2)a	26.4 (23.3 to 30.9)
Current smoker (%)	25 (8.3%)
Diabetes mellitus (%)	20 (6.6%)
Dyslipidemia (%)	78 (25.7%)
Hypertension (%)	95 (31.3%)
Statin use (%)	54 (17.8%)
Periodontal bone loss (% of populationa)	Edentulous (1.5%) None (0%) Mild (37.4%) Mild-to-moderate (1.1%) Moderate (30.5%) Moderate-to-severe (4.2%) Severe (25.2%)
Subsequent cardiovascular event (%)	13 (4.3%)
Framingham Risk Scoreb
<10	114 (82.0%)
10–19.9	21 (15.1%)
≥20	4 (2.9%)

Values are median (P25 to P75), or n (%).

^aAvailable in 295 subjects

^bAvailable in 139 subjects.

3.2. Periodontal disease activity independently associates with arterial inflammation

We previously reported a significant association between PD_inflammation and arterial inflammation in this cohort³¹. Herein, we additionally observe that the relationship between PD_inflammation and arterial inflammation remained significant after adjusting for CVD risk factors (age, sex, hypertension, diabetes, smoking, and dyslipidemia) (standardized β [95% CI]: (0.30 [0.20 to 0.40], P <0.001) or after additionally adjusting for statin use (0.30 [0.20 to 0.40], P <0.001). Additionally, there was a non-significant trend towards an association between PD_inflammation and CAC (r = 0.12, P = 0.06).

3.3. Periodontal disease activity independently associates with subsequent MACE

During follow-up, 13 participants in the outcomes study (4.2%) developed MACE, including four myocardial infarctions, one coronary revascularization, five strokes, one cardiovascular death, and two peripheral arterial disease events. PD_inflammation and arterial inflammation were higher among individuals with subsequent MACE (Fig. 2). Moreover, PD_inflammation predicted the risk of subsequent MACE (Table 2 and Fig. 3), yielding a 2.25-fold increased risk per SD increase in periodontal SUV in univariable analysis (standardized HR [95% CI]: 2.25 [1.47 to 3.44], P <0.001). Furthermore, the relationship between PD_inflammation and MACE remained significant after adjustments for: 1) risk factors shared by periodontal and CVD (i.e., age, sex, smoking, hypertension, and diabetes mellitus), 2) Framingham risk score, 3) preexisting atherosclerotic disease burden (i.e., CAC score), and 4) periodontal bone loss at baseline (Table 2). Further, PD_inflammation remained predictive of CVD after adjustment for history of prior cancer (standardized HR [95% CI]: 2.31 [1.49 to 3.58], P <0.0001).

TABLE 2.

Uni- and multivariable analysis of periodontal activity versus MACE

Model Predictor	Model Covariablesa	Model outcome
		MACE
		Standardized HR (95% CI)	P
Inflammation of the periodontium (PDinflammation) a	None	2.25 (1.47 to 3.44)	<0.001
	Age and sex	1.94 (1.31 to 2.87)	0.001
	Framingham Risk Scoreb	1.79 (1.16 to 2.75)	0.008
	Age, sex, smoking, HTN, and DM	1.93 (1.31 to 2.84)	0.001
	Statin use	2.09 (1.36 to 3.23)	0.001
	CAC	2.19 (1.41 to 3.39)	<0.001
	Periodontal bone lossc	3.16 (1.60 to 6.22)	0.001
	Arterial inflammationd	1.80 (1.12 to 2.89)	0.016

Continuous variables were standardized; a backward (conditional) model was used when more than one covariable was entered into the model.

^aThe predictor, Inflammation of the periodontium, was entered as a continuous variable.

^bFramingham Risk Score data for these sub-analyses are available in 106 individuals.

^cPeriodontal bone loss was evaluated on a five-point scale; the analysis excluded individuals who were edentulous.

^dArterial Inflammation is measured as target-to-background on positron emission tomography.

DM, diabetes mellitus; HTN, arterial hypertension; CAC, coronary artery calcium; HR, hazard ratio, CI, confidence interval.

FIGURE 3.

Kaplan-Meier survival curves showing MACE-free survival by periodontal activity. Survival curves relating PD_inflammation and MACE are shown. PD_inflammation was dichotomously characterized as “high” or “low” based on two distinct approaches: A) The first approach dichotomized PD_inflammation values by identifying which PD_inflammation threshold most accurately predicted future MACE events using receiver-operator curve (ROC) analyses. B) The second approach dichotomized PD_inflammation based on data distribution with a defined threshold SUV value ≥1 SD above the mean periodontal SUV. MACE, major adverse cardiovascular event; PD_inflammatlon, periodontal disease inflammation; ROC, receiver operating curve; SUV, standardized uptake value

Unlike the findings with PD_inflammation, periodontal bone loss did not associate with MACE, either in univariable analyses (1.24 [0.74 to 2.08], P = 0.405), or after adjustments for age, sex, smoking, hypertension, and diabetes (1.09 [0.66 to 1.82], P = 0.731) or for statin use (1.22 [0.72 to 2.02], P = 0.488).

3.4. Association between PD_inflammation and MACE events is mediated by arterial inflammation

We performed mediation (pathway) analyses to test the hypothesis that the relationship between PD_inflammation and MACE was significantly mediated by increased arterial inflammation. Arterial inflammation accounted for most of the relationship between PD_inflammation and MACE (standardized logs odds ratio, [95% CI]: 0.438 [0.019 to 0.880], P = 0.022), or 80% of the total effect (see Figure S1 in online Journal of Periodontology). This relationship (path) remained significant in analyses that adjusted for age and sex (P = 0.032), periodontal and CVD risk factors (i.e., age, sex, smoking, diabetes, arterial hypertension, dyslipidemia and family history of coronary artery disease; P = 0.036), as well as CAC and periodontal bone loss (P = 0.043). While unable to establish causality, this analysis delineates a significant pathway (PD_inflammation ⟶ arterial inflammation ⟶ MACE). On the other hand, the path that excludes arterial inflammation becomes non-significant (P = 0.579).

4 |. DISCUSSION

To our knowledge, this is the first advanced imaging study to link periodontal inflammation to subsequent major adverse cardiovascular events, showing that inflammation of the periodontium predicts MACE after correction for potential confounders and shared risk factors. Further, we observe that objectively measured inflammation of the periodontium (i.e., PD_inflammation) may represent a better predictor of MACE than measures of periodontal bone loss. Additionally, the current study suggests that the mechanism mediating the link between PD_inflammation and CVD may involve increased arterial inflammation.

While the relationship between periodontal disease and CVD has been previously examined,^5,8 important questions remained unanswered. Perhaps the most critical question pertains to the identification of the mechanistic pathway linking periodontal disease to CVD.² We previously observed an independent association between PD_inflammation and atherosclerotic inflammation that remained significant after robust adjustments for disease risk factors that are shared between PD and atherosclerotic diseases.^13,14 We posited that the association between PD and arterial inflammation may be particularly important, since arterial inflammation robustly associates with MACE risk.¹⁵ We hypothesized that PD associates with CVD events through a mechanism that involves increased arterial inflammation (Fig. 4). In support of this hypothesis, the findings of the current study demonstrate that: 1) PD_inflammation independently associates with both increased arterial inflammation as well as CVD after adjusting for risk factors for the diseases, and 2) mediation (pathway) analysis demonstrates that arterial inflammation significantly mediates the link between PD_inflammation and CVD, accounting for 80% of the total effect.

FIGURE 4.

Hypothesized pathway linking periodontal disease to CVD

The major cause of MACE is rupture of the lipid-rich atherosclerotic plaque that causes blockage of a major artery, resulting in an acute coronary syndrome.³² Established risk factors, including dyslipidemia, hypertension, diabetes mellitus, and smoking are associated with MACE. Nevertheless, >40% of all heart attacks and strokes occur in patients with “normal” cholesterol levels.³³ In humans, plaques at high-risk for rupture present distinct features, including intense inflammation and oxidative stress, large areas of necrosis composed of apoptotic cells with thinning, and inflammation of the protective fibrous cap.^34,35 There are significant data to suggest that local inflammatory foci, particularly in the oral cavity,^14,36,37 can impact the systemic inflammatory burden^38,39 and may contribute to the risk for MACE. Animal studies strongly suggest that local periodontal inflammation is associated with an increase in the incidence of atherothrombosis.^40–42 Furthermore, we previously reported, in this cohort, an association between inflammation of the periodontium and upregulated bone marrow leukopoiesis (an important source of inflammatory cells that traffic to the atherosclerotic milieu).³¹ Additionally, untreated, bacterially-induced periodontal inflammation may contribute to atherosclerotic plaque vulnerability and thrombosis by activating macrophages that elaborate enzymes including matrix metalloproteinases that degrade all components of the arterial extracellular matrix.^43,44 Inflammation can also increase fibrinogen-stimulated coagulation and accumulation of activated platelets at the site of thin fibrous cap or fissure.^45,46 Even low-grade inflammation may influence platelet activation potentially leading to platelet aggregation.⁴⁷ Platelet activity and aggregation has been associated with periodontitis in previous reports. Perumal et al.⁴⁸ demonstrated increase P-selectin surface expression on platelets and increased aggregation in periodontitis, as have others.^49,50 Moreover, treatment of periodontitis seems to reverse the propensity for platelet activation.^51,52

4.1 |. Limitations

The retrospective nature of this study limits the opportunity to compare the imaging findings with blood biomarkers, as these factors were variably assessed in the study population. Further, of the original cohort of imaged patients, 216 individuals either lacked PET coverage or had insufficient data quality to assess periodontal uptake of FDG (thus constraining the final sample size to 304 individuals). As a result, the number of observed clinical events was modest; only thirteen study participants developed MACE during the follow-up period. Notably, the current study participants were identified from a database of patients who had undergone ¹⁸F-FDG-PET/CT for clinical indications, mainly cancer screening. Further studies are needed to determine the generalizability of our findings to broader patient populations, especially since many risk factors are shared by malignancy and periodontal disease. We conscientiously included only patients that had documented cancer remission for at least 1 year. Additionally, the association between PD_inflammation and CVD events remained robust after adjusting for history of cancer, suggesting that this potential confounder is not responsible for our findings. Nonetheless, it will be important to replicate these findings in future, larger studies. Nevertheless, these limitations are counterbalanced by significant strengths, including the unique, simultaneous, multi-modality assessment of disparate tissues and their association with MACE in humans.

4.2 |. Future directions

This study should prompt further evaluation of the mechanisms linking PD_inflammation to CVD. These results underscore the intriguing hypothesis that improving oral health and aggressively treating PD_inflammation might reduce arterial inflammation and builds upon the observation that treatment of periodontal inflammation reduces systemic inflammation ⁵³ and improves endothelial function.⁵⁴ As such, these findings should stimulate studies to test whether treating PD_inflammation ameliorates arterial inflammation and reduces the risk of CVD events. Should future studies show such a benefit, PD_inflammation could ultimately be recognized as a modifiable atherosclerotic risk factor, and more aggressive screening and therapy for PD_inflammation could be pursued for individuals at higher risk for atherosclerotic diseases.

5 |. CONCLUSIONS

This study demonstrates that periodontal disease inflammation predicts the risk of MACE independently of risk factors for atherosclerosis or PD. The study may also shed light on the mechanism mediating that association, by suggesting that increased arterial inflammation links PD_inflammation to CVD. Accordingly, the current study suggests that PD_inflammation might represent an independent and potentially modifiable risk factor for atherosclerotic diseases and provides novel insights into the mechanism mediating that association.