Вы находитесь на странице: 1из 11

Periodontology 2000, Vol.

23, 2000, 110120 Copyright C Munksgaard 2000


Printed in Denmark All rights reserved
PERIODONTOLOGY 2000
ISSN 0906-6713

Oral disease, cardiovascular


disease and systemic inflammation
J AMES D. B ECK, G ARY S LADE & S TEVEN O FFENBACHER

Atherosclerosis has been defined as a progressive to occur in persons who are older, male, of lower
disease process that involves the large to medium educational status, have fewer financial resources,
sized muscular and the large elastic arteries. Athero- who smoke, are stressed, and are socially isolated.
sclerosis can lead to coronary heart disease, as well These commonalties hint that periodontal diseases
as myocardial and cerebral infarctions. Periodontitis and heart disease may share a similar causative
is an inflammatory reaction (to gram-negative, an- pathway.
aerobic bacterial infections) of the supportive tissues
surrounding the tooth, including the periodontal
ligament, cementum, and alveolar and supporting
bone. Associations between periodontal
World Health Organization statistics indicate that disease and atherosclerosis and
in 1995, cardiovascular diseases were responsible for coronary heart disease
20% of deaths worldwide, numbering 14 million
people. In developed countries, cardiovascular dis- A number of studies in the 1990s have shown associ-
eases account for 50% of deaths and are the prin- ations between oral conditions and atherosclerosis
ciple cause of death, while in developing countries and coronary heart disease. Other publications have
it ranks third, accounting for 16% of deaths (34). described and evaluated these associations in some
As shown in Table 1, heart disease is a major cause detail (2, 9, 17, 29), but they are summarized here
of death in the United States. Overall, in 1995 heart to provide a background from which to discuss the
diseases ranked at the number one killer, accounting possible mechanisms.
for more than 2 million deaths and a death rate of The findings from eight epidemiological studies
880 deaths per 100,000 people. However, the death (three case-control and five prospective studies) are
rates increase with age, rising to over 1800 deaths presented in Table 2. Mattila et al. (20) conducted
per 100,000 in the 65 years and over age group (27). two separate case-control studies that totaled 100
These two diseases appear to have a number of patients with acute myocardial infarction and 102
characteristics in common as they are more likely controls selected from the community at random.
Patients in the first series were 40 consecutive men
aged 50 years or less admitted because of acute myo-
cardial infarction and the controls were matched for
Table 1. Rank, deaths and death rates due to
age, sex and neighborhood from the community and
disease of the heart by age in 1995
invited for examination when the patient was admit-
Age group Rate per ted. The second series consisted of 60 consecutive
(years) Rank Number 100,000 population
cases, either men aged 60 years or less or women
All 1 2,312,203 880.0
aged 65 years or less. The dental examination was
14 5 256 1.6
performed while patients were in the hospital or
514 6 269 0.7
shortly thereafter. The dental index, which was used
1524 5 964 2.7
in all of Mattilas publications, is the sum of scores
2544 4 16,719 20.1
for the number of carious lesions, missing teeth,
4564 2 101,975 199.3
probing depth measures (including pus in the
65 1 627,844 1842.5
pocket), the number of periapical lesions and the
Source: National Center for Health Statistics (2).
presence or absence of pericoronitis. These index

110
Oral disease, cardiovascular disease and systemic inflammation

Table 2. Measure of association in epidemiological studies of periodontal disease and coronary heart
disease or stroke
Author/year Cases/controls Measure of association and
(reference) non-cases) Outcome Exposure 95% confidence interval
Case-control
Mattila 1989 (20) 100/102 Myocardial infarction Total Dental Index
Mattila 1993 (22) 100 Coronary atheromatosis Total Dental Index
Grau 1997 (10) 166/166 Stroke Total Dental Index
Prospective
Mattila 1995 (23) 52/(162) New myocardial infarction or Total Dental Index
death
DeStefano 1993 1786/(7974) Admits/death coronary heart Periodontal Index
(5) disease
Men 50
Joshipura 1996 New coronary heart disease Tooth loss due to
(14) periodontal disease
Beck 1996 (1) 203/(891) New coronary heart disease
58/(891) Fatal coronary heart disease
40/(911) Stroke
Genco 1997 (8) 68/(1304) New coronary heart disease Bone loss
---odds ratio; ---relative risk; ---incidence odds ratio; hazard ratio.

scores showed that patients had worse dental health ides and severe coronary atheromatosis for males.
than controls in both studies. Logistic regression These associations remained significant even though
analysis indicated that the association between poor total cholesterol, high-density lipoprotein chol-
dental health and coronary heart disease persisted esterol, smoking, hypertension, social class, and
after controlling for age, total cholesterol, high-den- body mass index were in the model and nonsignifi-
sity lipoprotein, triglycerides, C peptide, hyperten- cant.
sion, presence of diabetes, and smoking. An article by Grau et al. (10) looking at the re-
A second article by Mattila et al. (22) reports on lationship between infection and stroke does shed
associations between dental infections and athero- some light on the Total Dental Index used by Mattila.
sclerosis from a case control study (using the same In their case-control study, Grau et al. found that the
subjects as the earlier study). This second study was Total Dental Index was independently associated
limited to 100 subjects (88 men and 12 women) who with cerebrovascular ischemia (odds ratio, 2.6; 95%
were referred for diagnostic coronary angiography. confidence interval, 1.18 to 5.7). However, they
The left main coronary artery, the a, b and c portions found that only the periodontal components of the
of the right coronary artery, the circumflex artery, Total Dental Index were responsible for this associ-
and the left anterior descending artery were assessed ation.
and graded for degree of occlusion on a five-point The first prospective study conducted by Mattila
scale. A semiquantitative estimate of atherosclerotic et al. (23) is a 7-year follow-up of the cases from their
mass was obtained by multiplying the scores of long earlier study. They followed 182 men and 32 women
(5 mm) lesions by 2 and giving an additional score who had experienced a myocardial infarction in or-
of 1 to each segment with stenotic lesions in its pe- der to determine new fatal and non-fatal coronary
ripheral branches. Coronary atheromatosis scores events and overall mortality. The baseline dental
were the sum of each of the components. These measures were the Total Dental Index (which in-
scores were then divided into tertiles with the cluded caries, periodontitis, periapical lesions, and
highest tertile compared with the other two. The pericoronitis) and the Pantomography Index (num-
dental index score was a combined score for dental ber of vertical bone pockets, furcations and peri-
caries and periodontal infections. No significant as- apical lesions, and lesions caused by fourth-degree
sociations between dental infections and coronary caries or pericoronitis) used in Mattilas earlier
atheromatosis was found for the 12 women. In a studies and recorded at the time of first admission
multivariate analysis, significant associations were for a myocardial infarction. Cox proportional hazard
found between dental infections, age, and triglycer- models were developed for the 52 patients who met

111
Beck et al.

the outcome criteria. The Total Dental Index (coef- Beck et al. (1) conducted analyses on data from a
ficient0.18, SE0.06 was significant in a model that cohort study using combined data from the Norma-
included the Pantomography Index, the number of tive Aging Study and the Dental Longitudinal Study
previous infarctions, diabetes, body mass index, hy- sponsored by the Department of Veterans Affairs.
pertension, smoking, total cholesterol, high-density The Normative Aging Study (3) is a longitudinal
lipoprotein cholesterol, triglycerides, socioeconomic study of male veterans in the Boston area, who re-
status, gender and age. Our conversion of the Cox ceive their health care from community sources,
coefficients from the hazard models indicates that rather than VA facilities. The Dental Longitudinal
the hazard ratio for the Total Dental Index was 1.2. Study (15) was a substudy of the Normative Aging
Thus, for each unit increase in the Total Dental Index Study begun 7 years later and included only men
(range010), the hazard ratio for new coronary who were systemically healthy at the time of the
event increased by a factor of 1.2. dental examinations. Mean bone loss scores and
DeStefano et al. (5) investigated coronary heart worst probing pocket depth scores per tooth were
disease and mortality based upon the first US Na- measured on 1147 men during 19681971 and again
tional Health and Nutrition Examination Survey, during a follow-up examination 18 years later. As
which followed subjects for 14 years. This study shown in Table 2, incidence odds ratios adjusted for
examined several potentially confounding variables age and established cardiovascular risk factors were
including age, gender, race, education, marital state, 1.5, 1.9, and 2.8 for bone loss and total coronary
systolic blood pressure, total cholesterol levels, body heart disease, fatal coronary heart disease and
mass index, diabetes, physical activity, alcohol con- stroke, respectively.
sumption, poverty index and cigarette smoking. De- When the degree or severity of the exposure is re-
Stefano et al. demonstrated that, among the nearly lated to the incidence of the disease, we consider
10,000 subjects analyzed, those with periodontitis this to be important information in establishing that
had a 25% increased risk of coronary heart disease exposure as a cause. This dose-response effect or
relative to those with minimal periodontal disease, biological gradient is important because, if a factor
adjusted for the covariables mentioned above. In is of causal importance, then the risk of developing
men younger than 50 years of age, periodontal dis- the disease should be related to the degree of ex-
ease had an effect on coronary heart disease inci- posure to the factor.
dence with a relative risk of 1.72. The Beck et al. study (see above and reference 1)
Joshipura et al. (14) followed 44,119 men who reported the level of bone loss in relation to the
were health professionals (who had no reported cumulative incidence of coronary heart disease, fatal
coronary heart disease symptoms at baseline, plaus- coronary heart disease and stroke respectively. For
ible dietary histories and no missing information on all three outcomes, the point estimates and standard
age or number of teeth) for 6 years. All information errors show that increasing levels of bone loss at
was obtained from mailed questionnaires. Analyses baseline are accompanied by a higher cumulative
were conducted for all study subjects and separately frequency of occurrence of disease 18 years later, in-
for dentists, who comprised 58% of the subjects, and dicating a biological gradient between exposure and
the results were similar. The authors reported no as- occurrence of disease.
sociation between self-reported history of peri- Genco et al. (8) investigated the association be-
odontal disease with bone loss and coronary heart tween periodontal infection and risk of cardiovascu-
disease with a relative risk of 1.04 (95% confidence lar disease in 1372 Native Americans of the Gila River
interval0.86, 1.25) when controlling for age, body Indian Community, a group with a high prevalence
mass, exercise, smoking habits, alcohol consump- of diabetes mellitus. At baseline alveolar bone level
tion, family history of myocardial infarction before was assessed and cardiovascular status was moni-
age 60 and vitamin E consumption. However, men tored for up to 10 years, using electrocardiograms.
who had 010 teeth compared with men with 25 or New cardiovascular disease occurred in 68 people.
more teeth had a relative risk of 1.40 (95% confi- Among all age groups, bone level was predictive of
dence interval1.04, 1.87). Stratification of tooth loss cardiovascular disease but did not remain significant
by periodontal history showed that the association in a multivariate analysis. However, for persons 60
was limited to men with periodontal disease. Among years bone level was predictive of cardiovascular dis-
6619 men with periodontal disease, the multivariate ease, with an odds ratio of 2.68, (95% confidence in-
relative risk was 1.67 (95% confidence interval1.03, terval1.30, 5.5), adjusting for the effects of gender
2.71) for men with 10 or fewer remaining teeth. and duration (10 years) of diabetes. Since this group

112
Oral disease, cardiovascular disease and systemic inflammation

either did not smoke or had very low smoking levels, vascular risk, including rheological indicators of
the study indicates that the bone level cardiovascu- hypercoagulation. Mattila et al. (27) conducted a
lar disease association was not due to smoking, a risk study on the relationship between dental infections
factor for both conditions. and several serum factors (factor VIII, factor VIII ris-
In reviewing the evidence for oral conditions as tocetin cofactor activity and von Willebrand factor).
risk factors for heart disease, one is struck by the They studied 40 consecutive patients with acute
consistency of the associations across studies that myocardial infarction and 41 randomly selected con-
used a variety of measures for both the exposure and trols. All patients and controls were males aged 50
the outcome (Table 2). The strength of the associ- years and over. Blood samples for the assays were
ations are moderate, with most ranging from 1.2 to taken for patients upon arrival at the hospital and at
1.5 and only the adjusted incidence odds ratios for 1, 4, and 12 weeks and for controls at entry into the
fatal heart disease and stroke in the Beck et al. study study. Dental status was measured by the Total Den-
were about 2.0, with the incidence odds ratios for tal Index (described previously). These factors were
extent of probing depth greater than 40% and total compared in persons with dental status scores below
coronary heart disease approaching 3.0. Yet, a factor and above the median Total Dental Index score. The
associated with even 20% of the risk for coronary mean scores of all the factors were higher among pa-
heart disease would be associated with a very large tients and controls with poorer dental status than
number of cases and is worthy of attention. It should among patients and controls with better dental sta-
also be noted that all of these studies controlled for tus, but only the von Willebrand factor scores were
relevant known risk factors for heart disease that statistically significant among controls (P0.02). In
were available to the investigators (Table 3) but, as a similar analysis using the dental Pantomography
in most studies, uncontrolled and residual con- Index, similar trends were found, but both factor VIII
founding could still exist. ristocetin cofactor activity score (P0.01) and the
von Willebrand factor scores (P0.05) were signifi-
cantly higher among controls with poorer dental sta-
Associations between periodontitis tus. While patient vs control differences were not re-
and cardiovascular, hematological ported, it appears that there were no significant dif-
and rheological risk markers ferences between serum factor scores between
patients and controls, although the trends indicated
In addition to studies that link periodontal disease higher scores among patients than among controls.
to cardiovascular outcomes, there is literature that Thus, relatively healthy controls with poor dental
begins to indicate that periodontitis is also associ- status did have higher von Willebrand scores (a
ated with changes in blood cellular and biochemical marker of endothelial damage) than controls with
composition that are secondary markers of cardio- better dental status, but no differences were shown

Table 3. Degree of adjustment for associations between periodontal disease and coronary heart dis-
ease and stroke
Case-control studies Adjusted for
Mattila 1989 (20) High-density lipoprotein, smoking, hypertension, age, triglycerides, social class, diabetes,
total cholesterol and C peptide
Mattila 1993 (22) Age, triglycerides, high-density lipoprotein, smoking, hypertension, social class and body mass
index
Grau 1997 (10) Diabetes mellitus, pre-existing vascular diseases, social class and smoking
Prospective studies Adjusted for
Mattila 1995 (23) Smoking, hypertension, age, gender, triglycerides, social class, diabetes, serum lipids,
body mass index and previous myocardial infarctions
De Stefano 1993 (5) Age, gender, race, education, poverty, marital status, systolic blood pressure, cholesterol, diabetes,
body mass index, physical activity, alcohol and smoking
Joshipura 1996 (14) Age, body mass index, exercise, smoking, alcohol, vitamin E intake, family history of myocardial
infarction before age 60 years
Beck 1996 (1) Age, body mass index, total cholesterol, systolic blood pressure, diastolic blood pressure,
low-density lipoprotein cholesterol and smoking
Genco 1997 (8) Gender and duration of diabetes

113
Beck et al.

among patients who had just suffered an acute myo- mild to moderate gingivitis with no probing pocket
cardial infarction. depths 4 mm and no radiographic evidence of al-
Kweider et al. (16) compared fibrinogen and white veolar bone loss. The purpose of the study was to
cell count in patients with gingivitis or periodontitis determine the relationship between periodontal sta-
and in periodontally healthy controls. Fifty consecu- tus and both C-reactive protein and haptoglobin
tive gingivitis or periodontitis patients aged 2550 levels at baseline, to follow the C-reactive protein
years attending the dental hospital and 50 individ- and haptoglobin levels over time, and to assess any
uals with relatively healthy periodontal tissues (hos- changes in C-reactive protein and haptoglobin sub-
pital staff and patients) were evaluated using the sequent to both local and then systemic periodontal
Plaque Index, Gingival Index, and the Community therapy. Each acute-phase reactant was significantly
Periodontal Index of Treatment Needs. Venus blood increased in serum from adult periodontitis patients
was sampled for blind measurement of fibrinogen compared with controls (P0.001 for each). In ad-
and white cell count. Bivariate analyses showed that dition, they found that C-reactive protein and hapto-
periodontal patients had significantly higher fi- globin were significantly elevated in patients with
brinogen (g/l) and white cell counts (109/l) than the most disease active sites over a 6-month period
controls (P0.001 for both). Multivariate analyses (P0.02 and P0.001 respectively). Haptoglobin
showed an association between the gingival index levels declined after scaling and root planing
and both fibrinogen and white cell count. Each of (P0.01) and after a 2-year administration of a non-
these associations was independent of age, smoking steroidal anti-inflammatory drug, significantly de-
status, social class, and group status (patient vs con- creased haptoglobin levels were noted (P0.005). C-
trol). The Plaque Index and Community Periodontal reactive protein levels declined by 3540% after 12
Index of Treatment Needs were reported to show years of treatment with the drug (P0.05).
similar trends, but were not significant. In this study,
it appeared that individuals with poorer Gingival
Index scores had higher fibrinogen scores and white Potential biological mechanisms
cell counts, irrespective of whether they had been
classified as periodontal patients or controls. Infection has been recognized as a risk factor for
Ebersole et al. (7) studied a group of 40 adult peri- atherogenesis and thromboembolic events (5, 18, 20,
odontitis patients aged 3555 years both cross-sec- 24, 28, 32). Gram-negative bacteria or the associated
tionally and longitudinally. These patients had no lipopolysaccharide (endotoxin), when presented as
uncontrolled medical conditions or abnormal blood a systemic challenge in animal models, can induce
chemistries and no history of frequent periodontal inflammatory cell infiltration into major blood ves-
abscesses. The 35 controls used for the cross-sec- sels, vascular smooth muscle proliferation and vas-
tional part of the study were aged 3563 years, and cular fatty degeneration and intravascular coagu-
to be eligible for the study, they could only exhibit lation. The remarkable similarities of bacterially in-
duced vascular pathology and the natural history of
atherogenesis has led certain investigators to suggest
that, in addition to genetic, lifestyle and dietary in-
fluences, infections of unknown origin may contrib-
ute to the observed cardiovascular pathology. The
chronic inflammatory burden of periodontal infec-
tion and the host response provide the basis for our
hypothetical model of the observed associations be-
tween periodontal disease and atherosclerosis, coro-
nary heart disease and stroke. This conceptual
model is presented in Fig. 1 and described more fully
in our 1996 article (1). Fundamentally, this model
emphasizes that among certain individuals there
may be an underlying hyperinflammatory trait in re-
sponse to stimuli that is manifest by an excessive
Fig. 1. Proposed model. Source: Beck et al. (1). LPS:
lipopolysaccharide. PGE2: prostaglandin E2. IL-1b: in-
production of pro-inflammatory cytokines and lipid
terleukin 1b. TNFa: tumor necrosis factor a. TxB2: mediators by monocytes and other cell types. This
thromboxane B2. hyperinflammatory trait (M) may be induced by

114
Oral disease, cardiovascular disease and systemic inflammation

genetic, behavioral and environmental exposure and probing pocket depth, both recorded in millimeters
may serve as a common antecedent to both cardio- using the National Institute of Dental Research
vascular and periodontal risk. However, in the pres- probe that has 2-mm colored bands. For these meas-
ence of a pathogenic oral flora and resultant bactere- urements, all fractional millimeter measurements
mias and systemic inflammation, this may provide were rounded down to the nearest whole millimeter.
an added burden to the host increasing the preva- Clinical attachment level was then calculated by the
lence and severity of cardiovascular disease. Thus, computerized data entry system for all sites with
since coronary heart disease is a multifactorial dis- valid measurements of both recession and probing
ease process, we envision that periodontitis may pocket depth. Periodontal measurements were made
contribute to a certain fraction of cases, but that in two randomly selected quadrants (one quadrant
many other factors can play an independent causa- in the maxilla and one quadrant in the mandible) for
tive role in atherogenesis and infarctive disease. all teeth other than third molars, partially erupted
Further details of this model that emphasize the ob- teeth or retained roots.
ligate duality of both microbial exposure and in- Blood samples were collected from persons aged 1
flammatory responses in the proposed linkage be- year or more using venipuncture in volumes varying
tween periodontitis and coronary heart disease are from 7 ml for children aged 13 years to over 100 ml
described elsewhere (1). In the present investigation in some adults aged 2059 years. Serum was separ-
we sought to determine whether periodontitis was ated using standard laboratory techniques, and a
associated with an acute phase response, as sug- minimum volume of 0.3 ml was required for assess-
gested by the earlier by the report of Ebersole (7) and ment of C-reactive protein, which was quantified by
as depicted on the right hand side of our working latex-enhanced nephelometry using a fully auto-
model. To test this hypothesis we began by exam- mated Behring Nephelometer Analyzer System (11).
ining a nationwide database focussing on the acute- In summary, diluted, centrifuged samples of serum
phase response and C-reactive protein, specifically. were mixed with latex particles coated with rabbit
antiC-reactive protein antibodies. An accelerator
reagent containing detergent was added to the reac-
Recent evidence from United States tion mixture to enhance binding of the antigen-anti-
body complex. After 6 minutes, light scattering,
national survey data (NHANES III) which is proportional to the concentration of the an-
alyte present in the sample, was measured by a
Methods
nephelometric procedure. An automatic blank sub-
The third National Health and Nutrition Examina- traction was performed, and C-reactive protein con-
tion Survey (NHANES III) aimed to provide national centrations in mg/dl were calculated with reference
estimates of the health and nutritional status of the to a calibration curve. Using this method, the mini-
United States civilian, noninstitutionalized popula- mum detectable concentration of C-reactive protein
tion aged 2 months and older. The survey was con- was 0.21 mg/dl. Repeat testing was done on all
ducted in two phases, from 19881991 and 1992 samples with concentrations exceeding 1.0 mg/dl.
1994. Each phase, and the entire survey, provides na- Some 39,695 people were selected for the two
tional probability estimates for a broad range of self- phases of NHANES III, of whom 33,994 (86%) were
reported, examiner-assessed and laboratory-quan- interviewed in their homes. All interviewed people
tified indicators of health and nutritional status. Full were invited to mobile examination centers for a
documentation of the survey has been provided medical (including dental) examination and collec-
elsewhere, and descriptive findings on oral health tion of biological samples, conducted within 4 weeks
status from the first phase have been reported (6). In of the interview. Seventy-eight percent (30,818) of
summary, the design of NHANES III was a complex, the selected people attended the mobile examina-
multi-stage, stratified, clustered sample survey of the tion center. Some 28,059 people had dental exami-
civilian, noninstitutionalized population in the nations although 8786 of them were excluded from
United States. Sampled people were interviewed in periodontal assessment because they were aged less
their homes and were invited to attend a mobile ex- than 12 years, 1957 were excluded because they were
amination center for a standardized medical (includ- edentulous, 1165 were excluded because of medical
ing dental) examination, physical measurements contraindications and 254 had no periodontal as-
and collection of biological samples. The periodontal sessment for unspecified reasons. Of the remaining
assessment included measurement of recession and 15,897 dentate people who had a periodontal exami-

115
Beck et al.

nation, C-reactive protein data were available for variates were assessed for potential confounding of
14,979. This analysis was confined to the 12,949 of the periodontalC-reactive protein relationship: age
these individuals who were aged 18 years or more. (seven categories) and self-reported smoking history
Younger subjects were excluded in an attempt to ex- (never smoked, former smoker and current smoker).
clude false pockets, which were expected to distort Differences in mean C-reactive protein levels
relationships between periodontal infection (as among three groups defined by extent of periodontal
measured by pocket depth) and systemic C-reactive pocketing were evaluated using t-tests, both for cru-
protein levels. de differences and after direct age standardization
For this analysis, unit record data were obtained among the three groups. The standard population
from public-release CD-ROM ASCII files provided by used was the United States population recorded at
the National Center for Health Statistics. Data were the 1980 census and categorized into 10-year age
read from separate files containing interview data (to categories. Standardization was achieved using the
obtain demographic and smoking history), dental stdvar/stdwgt options in SUDAANs PROC
examinations and laboratory assays. Variables desig- DESCRIPT (30). Stratified analyses were then con-
nating the sampling design and unit record weights ducted among three categories of smoking. For the
were obtained from the examination file. The sample analysis of three C-reactive protein categories, the
weights permit analysis to be generalized to the chi-square statistic was used to evaluate crude dif-
United States population by correcting for: 1) differ- ferences in proportions among groups defined by
ent probabilities of subject selection that were in- the extent of periodontal pocketing. Stratified analy-
herent in the sampling design; and 2) different levels ses, controlling for age and cigarette smoking, were
of nonresponse among sociodemographic groups. evaluated using the Mantel-Haenszel statistic. All
For this analysis, the dependent variable was sys- these analysis were conducted in SUDAAN using the
temic C-reactive protein level, which was used both documented design and weight options specified in
as a continuous variable and in three ordinal cate- the NHANES III documentation.
gories (0.21 mg/dl, which is below the detectable
threshold, 0.210.99 mg/dl and 1.0 mg/dl or more).
This highest category has been used as evidence of Results
clinically meaningful infection, as reported by Tracy
et al. (19, 33). When used as a continuous variable, Table 4 presents the relationship between the per-
values below the detectable threshold were imputed cent of periodontal sites with a probing depth of 4
to 0.10 mg/dl approximately halfway between zero mm and C-reactive protein levels. Age-standardized
and the 0.21 mg/dl threshold. The main exposure C-reactive protein levels are presented in two forms,
variable was the extent of periodontal pocketing at mean C-reactive protein, the percent of persons with
the 4 mm threshold as described by Carlos et al. a clinically meaningful C-reactive protein level of 1
(4) Extent scores represent the percentage of probed mg/dl. Both categorizations of C-reactive protein
sites that have pocketing of at least 4 mm, and this values appeared to be related to the extent of sites
was categorized into three levels (0%, 110% and with probing depth of 4 mm. In fact, the mean
10% of sites) to provide an ordinal indicator of ex- scores appeared to present a monotonic association.
posure to periodontal infection. Two additional co- People with 10% or more of their sites having a prob-

Table 4. NHANES III: percentage of sites with 4 mm of periodontal probing depth and C-reactive
protein measures in people aged 8 years
Age-standardized mean Age-standardized % of people
C-reactive protein with 1 mg/dl C-reactive
concentration (mg/dl) protein
% of sites with No. of persons in
4 mm probing depths sample (unweighted) Mean (SE) % (SE)
0% 9145 0.29 (0.01) 6.0 (0.37)
19% 2253 0.31 (0.02) 5.5 (0.69)
10% 1550 0.41 (0.04)aa,bb 12.5 (2.27)aa,bb
Everyone 12,948 0.30 (0.01) 6.3 (0.36)
aa
Differs significantly from 0% severity group, P0.01.
bb
Differs significantly from 19% severity group, P0.01.

116
Oral disease, cardiovascular disease and systemic inflammation

ing depth of 4 mm have significantly higher C-re-


active protein levels than those with no 4 mm
probing depths and those with 19% of their sites
having 4 mm probing depths. However, subjects in
the lowest two probing depth groups did not have
significantly different C-reactive protein levels.
Since smoking is known to be associated with
both periodontal disease and C-reactive protein
levels, associations between periodontal disease and
C-reactive protein, stratified by smoking status, are
presented in Table 5. Never smokers present essenti-
ally the same positive association between probing Fig. 2. NHANES III: age-specific C-reactive protein levels
and extent of probing depths of 4 mm for those who
depths of 4 mm and C-reactive protein levels that
have never smoked
were seen in Table 4. In former smokers, the highest
C-reactive protein levels appeared in those persons
with 19% of their sites with a probing depth of 4
mm. In current smokers the 10% group has signifi-
cantly higher C-reactive protein levels than the 19%
group, but they were not significantly greater than
those with no probing depths of 4 mm. Thus, C-
reactive protein levels appeared to be higher in non-
smokers with the greatest extent of probing depths,
but a current or past smoking history weakens the
periodontal diseaseC-reactive protein association.
On the other hand, the smoking-C-reactive protein
relationship only is evident for those with no prob-
ing depths 4 mm, indicating that the smoking-C-
reactive protein relationship may only be valid in
people without periodontal disease.
Fig. 2 and 3 represent age groupspecific associ- Fig. 3. NHANES III: age-specific C-reactive protein levels
ations between dichotomized probing depth extent and extent of probing depths of 4 mm for those who
currently smoke
scores and C-reactive protein levels for never
smokers and current smokers respectively. For never

Table 5. NHANES III: percentage of sites with 4 mm of periodontal probing depth and C-reactive
protein measures in people aged 18 years by smoking status
Age-standardized mean Age-standardized % of persons
C-reactive protein with 1 mg/dl C-reactive protein
concentration (mg/dl)
% of sites with No. of persons in
4 mm probing depths sample (unweighted) Mean (SE) Mean (SE)
Never smokers
0% 5089 0.28 (0.01) 5.4 (0.47)
19% 1092 0.27 (0.02) 4.5 (0.79)
10% 602 0.45 (0.06)aa, bb 14.0 (2.46)aa, bb
Former smokers
0% 1995 0.27 (0.02) 4.6 (0.66)
19% 482 0.40 (0.05)a 9.4 (3.49)
10% 368 0.34 (0.05) 5.2 (1.30)
Current smokers
0% 2061 0.36 (0.03) 9.1 (1.18)
19% 679 0.28 (0.03) 3.3 (0.74)a
10% 580 0.43 (0.06)b 16.0 (4.70)b
aa
Differs significantly from 0% severity group, P0.01.
bb
Differs significantly from 19% severity group, P0.01.
a
Differs significantly from 0% severity group, P0.05.
b
Differs significantly from 19% severity group, P0.05.

117
Beck et al.

smokers, it appeared that 1.0 mg/dl C-reactive pro- flammation. It should be pointed out, however, that
tein levels were more frequent in people with prob- C-reactive protein is a nonspecific marker of the
ing depth scores 10% in all age groups and the acute-phase response. That is, many potential stim-
largest probing depth effects were seen in the age uli, including other infections, inflammatory con-
groups 2534 and 5564 years old. For the current ditions and trauma may also account for mild in-
smokers, the age-group specific pattern was mixed creases in C-reactive protein. However, in a com-
with the age groups 3544 and 65 years, presenting panion publication, we further report that the
a negative association between periodontal disease potential effects of these other possible causes for C-
scores and C-reactive protein levels. reactive protein elevation, such as arthritis and dia-
betes, do not appear to negate the effects of peri-
odontitis on increasing C-reactive protein levels (31).
Discussion Two recent articles by Ridker et al. (25, 26) add to
the accumulating evidence that inflammation may
During the 1990s, three case-control studies and five be causally related to clinical cardiovascular disease.
longitudinal studies have consistently shown associ- In a nested case-control study of over 14,000 men
ations between periodontal status and coronary apparently healthy at baseline, moderately elevated
heart disease or stroke. These associations, while plasma concentrations of C-reactive protein (a
weak to moderate in strength, have been demon- marker of inflammation) were associated with
strated across diverse populations and appear to be greater risk of future myocardial infarction, stroke,
independent of traditional risk factors as measured and symptomatic peripheral arterial disease. Aspirin
in these studies, and one study has demonstrated a therapy was related to a reduction in coronary heart
dose-response effect. Thus, clarifying the potential disease events as well as reduced levels of C-reactive
mechanisms behind these associations is of great in- protein. These findings imply that anti-inflamma-
terest. Since the earliest reported study by Mattila tory agents may have preventive benefits for cardio-
(20), this periodontal diseasecoronary heart disease vascular outcomes. Although the inflammatory stim-
association has been cited as evidence of the infec- uli responsible for the moderately increased C-reac-
tion hypothesis; that infections are a major cause of tive protein levels could not be unequivocally
atherogenesis and thromboembolic processes. Infec- determined in the subjects studied by Ridker et al.,
tious diseases require both a pathogen and a suscep- the study by Ebersole (7) and the present investiga-
tible host, and the expression of that disease depend tion suggest that periodontal infections may be one
both on the virulence of the pathogen and the im- important stimulus for mildly elevated C-reactive
mune response of the host to that pathogen. protein.
We presented a working model of the periodontal Finally, an increasing body of evidence suggests
diseaseatherosclerosis association (Fig. 1), which that some periodontal microbes are capable of trig-
hypothesizes that the presence of a M phenotype gering thromboembolic events. The potential sig-
places certain individuals at risk for both athero- nificance of oral gram-positive bacteria including
sclerosis and coronary heart disease and peri- streptococcal species, such as Streptococcus sanguis,
odontitis. Also, we suggest that certain factors, such in mediating thromboembolic events has been the
as diet, may exacerbate the hyperinflammatory research focus of Herzberg et al. (13) and is pertinent
monocyte phenotype and may thereby contribute to to our principal hypotheses. Systemic hematogenous
the morbidity of atherosclerosis and periodontal dis- exposure to oral streptococcus has been recognized
ease status. Furthermore, if a M individual con- as a potential initiator of bacterial endocarditis. Early
tracts periodontitis, the likelihood of atherogenesis immunization experiments with oral streptococcal
and thromboembolic events increases. Finally, peri- organisms (especially Streptococcus mutans for vac-
odontal infections may directly contribute to the cination against caries) resulted in the production
pathogenesis of atherosclerosis and thrombo- of autoimmune antivascular and anticardiac cross-
embolic events by providing repeated systemic vas- reactive immunoglobulins. Recently, landmark
cular challenges with the lipopolysaccharide, in- studies by Herzberg et al. have identified the cross-
flammatory cytokines and the microbes themselves. reactive immunodeterminants responsible (13). S.
In this report we provide further evidence that peri- sanguis contains an outer membraneassociated
odontitis elicits a sufficient vascular challenge to protein with the 7-mer sequence Pro-Gly-Glu-Gln-
trigger a mild acute-phase response with an increase Gly-Pro-Lys, which is identical to the platelet-inter-
in C-reactive protein levels, a marker of systemic in- active domain of type I and type III collagens. This

118
Oral disease, cardiovascular disease and systemic inflammation

bacterial protein thus presents or mimics the nor- addition, the present demonstration that peri-
mal platelet receptor that initiates thrombus forma- odontitis is a mild activator of the acute-phase re-
tion. Type III collagen is part of the wall and base- sponse is consistent with our earlier hypotheses that
ment membrane of vessels and is normally covered periodontitis may elicit a systemic inflammatory re-
with endothelial cells that mask these platelet bind- sponse. Since systemic hepatic, hematical and vas-
ing receptor sequences from circulating blood cells. cular inflammation is a classic hallmark of cardio-
Trauma or other noxious stimuli can expose these vascular disease, the possibility that periodontitis
receptors and thereby provide an important signal to may be one potential evoking stimulus for that sys-
initiate hemostasis or thrombosis. Herzberg et al. temic inflammatory response bears further con-
have repeatedly suggested the oral organisms such sideration.
as S. sanguis can present a similar thrombotic trig-
ger. Herzberg et al. also report that Porphyromonas
gingivalis (a gram-negative periodontal pathogen)
possesses properties similar to that of S. sanguis (13).
References
Finally, Haraszthy et al. (12) report that 19 of 27
1. Beck JD, Garcia RG, Heiss G, Vokonas P, Offenbacher S.
atheromas obtained from patients during endart- Periodontal disease and cardiovascular disease. J Peri-
erectomy were positive for bacterial DNA from peri- odontol 1996: 67(suppl): 11231137.
odontal pathogens. Of the 19, six were positive for 2. Beck JD, Offenbacher, S, Williams RW, Gibbs P and Garcia
Actinobacillus actinomycetemcomitans, six for P. gin- RAR. Periodontitis: a risk factor for coronary heart disease?
Ann Periodontol 1998: 3: 126141.
givalis, and seven for Prevotella intermedia. Taken
3. Bell B, Rose CL, Damon A. The Veterans Administration
together, these findings suggest that periodontal longitudinal study of health and aging. Gerontologist 1966:
pathogens are found systemically and may serve as 6: 179184.
a thromboembolic trigger. 4. Carlos J, Wolfe M, Kingman A. The extent and severity
In summary, it is our hypothesis that the presence index: a simple method for use in epidemiologic studies of
periodontal disease. J Clin Periodontol 1986: 13: 500504.
of a M phenotype places certain individuals at
5. DeStefano F, Anda RF, Kahn HS, Williamson DF, Russell CM.
risk for both cardiovascular disease and peri- Dental disease and risk of coronary heart disease and mor-
odontitis. Furthermore, we suggest that certain fac- tality. BMJ 1993: 306: 688691.
tors, such as diet or genetics, may exacerbate the 6. Drury TF, et al. An overview of the oral health component
hyperinflammatory monocyte phenotype and there- of the 19881991 National Health and Nutrition Examina-
tion Survey (NHANES III-Phase 1). J Dent Res 1996: 75(spec
by contribute to the morbidity of atherosclerosis and
issue): 620630.
periodontal diseases. Thus, the M individual may 7. Ebersole J, Machen R, Steffen M, Willmann D. Systemic
be at increased risk for developing atherosclerotic acute-phase reactants, C-reactive protein and haptoglobin
disease (left side of Fig. 1), especially if other cardio- in adult periodontitis. Clin Exp Immunol 1997: 107: 347
vascular disease risk factors are present. If peri- 352.
8. Genco R, Chadda S, Grossi S, Dunford R, Taylor G, Knowler
odontal pathogens are also present and periodontitis
W, Pettitt D. Periodontal disease is a predictor of cardio-
results, the host response of the M individual is vascular disease in a Native American population. J Dent
likely to be unregulated (right side of Fig. 1). Thus, Res 1997: (suppl): abstr.
periodontal infections may directly contribute to 9. Genco RJ. Periodontal disease and risk for myocardial in-
atherogenesis and thromboembolic events by pro- farction and cardiovascular disease. Cardiovasc Res 1998:
March: 3440.
viding repeated systemic vascular challenges of pro-
10. Grau AJ, Buggle F, Siegler C. Association between acute
inflammatory cytokines, periodontal pathogens, and cerebrovascular ischemia and chronic and recurrent infec-
lipopolysaccharide. The pro-inflammatory cytokines tion. Stroke 1997: 28: 17241729.
may be more closely associated with the more 11. Gunter EW, Lewis BG, Koncikowski SM. 1996 laboratory
chronic aspects of cardiovascular disease, such as procedures used for the Third National Health and Nu-
trition Examination Survey (NHANES III), 19881994. At-
formation and progression of arterial plaques, while
lanta, GA: U.S. Department of Health and Human Services
the microbial and lipopolysaccharide exposures may Public Health Service Centers for Disease Control and Pre-
be associated with the more acute aspects of coro- vention National Center for Environmental Health, 1996:
nary heart disease and stroke, such as thrombus for- 303413724.
mation. Validating these possible mechanisms will 12. Haraszthy VI, Zambon JJ, Trevisan M, Shah R, Zeid M, Gen-
co RJ. Identification of pathogens in athromatous plaques.
require considerable study in coming years, but
J Dent Res 1998: 77(IADR abstr): 666.
some early critical linkages between periodontitis 13. Hertzberg M, MacFarlane G, Liu P, Erickson P. The platelet
and cardiovascular disease appear to be gaining ad- as an inflammatory cell in periodontal diseases: interac-
ditional confirmatory support with further study. In tions with Porphyromonas gingivalis. In: Genco R, ed. Mol-

119
Beck et al.

ecular pathogenesis of periodontal disease. Washington, 25. Ridker P, Cushman M, Stampfer M, Tracy R, Hennekens C.
DC: Am Soc Microbiol, 1994: 247255. Inflammation, aspirin, and the risk of cardiovascular dis-
14. Joshipura KJ, Rimm EB, Douglass CW, Trichopoulos D, As- ease in apparently healthy men. N Engl J Med 1997: 336:
cherio A, Willett WC. Poor oral health and coronary heart 973979.
disease. J Dent Res 1996: 75: 16311636. 26. Ridker PM, Cushman M, Stampfer MJ, Tracy RP, Hennekens
15. Kapur K, Glass R, Loftus E, Alman J, Feller R. The Veterans CH. Plasma concentration of C-reactive protein and risk of
Administration longitudinal study of oral health and dis- developing peripheral vascular disease. Circulation 1998:
ease. Aging Hum Dev 1972: 3: 125137. 97: 425428.
16. Kweider M, Lowe GDO, Murray GD, Kinane DF, McGowan 27. Rosenberg HM, Ventura SJ, Maurer JD, Jenkins D, Kaster C.
DA. Dental disease, fibrinogen and white cell count: links Births and deaths: United States, 1995. Monthly vital stat-
with myocardial infarction? Scot Med J 1993: 38: 7374. istics report; Vol. 45, no. 3, suppl 2, p. 31. Hyattsville, MD:
17. Loesche WJ, Schork A, Terpenning MS et al. Assessing the National Center for Health Statistics, 1996.
relationship between dental disease and coronary heart 28. Saikku P, Leinonen M, Tenkanen L, Linnanmaki E, Ekman
disease in elderly U.S. veterans. J Am Dent Assoc 1998: 129: M, Manninen V. Chronic Chlamydia pneumoniae infection
301311. as a risk factor for coronary heart disease in the Helsinki
18. Lopes-Virella MF, Virella G. Immunological and microbio- heart study. Ann Intern Med 1992: 116: 273278.
logical factors in the pathogenesis of atherosclerosis. Clin 29. Seymour RA, Steele JG. Is there a link between periodontal
Immunol Immunopathol 1985: 37: 377386. disease and coronary heart disease? Br Dent J 1998: 184:
19. Macy EM, Hayes TE, Tracy RP. Variability in the measure- 3338.
ment of C-reactive protein in healthy subjects: implications 30. Shah BV, Barnwell BG, Bieler. SUDAAN users manual. Re-
for reference intervals and epidemiological applications. lease 7.0. Research Triangle Park, NC: Research Triangle In-
Clin Chem 1997: 43: 5258. stitute, 1996.
20. Mattila K, Nieminen M, Valtonen V et al. Association be- 31. Slade GD, Offenbacher S, Beck JD, Heiss G, Pankow JS.
tween dental health and acute myocardial infarction. Br Acute-phase inflammatory response to periodontal disease
Med J 1989: 298: 779782. in the US population. J Dent Res 2000: 79: 4957.
21. Mattila K, Rasi V, Nieminen M et al. von Willebrand factor 32. Thom DH, Grayston JT, Siscovick DS, Wang S, Weiss NS,
antigen and dental infections. Thromb Res 1989: 56: 325 Daling JR. Association of prior infection with Chlamydia
329. pneumoniae and angiographically demonstrated coronary
22. Mattila K, Valle MS, Nieminen MS, Valtonen VV, Hietaniemi artery disease. JAMA 1992: 268: 6872.
KL. Dental infections and coronary atherosclerosis. Athero- 33. Tracy RP, Psaty BM, Macy E, Bovill EG, Cushman M, Cor-
sclerosis 1993: 103: 205211. nell ES, Kuller LH. Lifetime smoking exposure affects the
23. Mattila KJ, Valtonen VV, Nieminen M, Huttunen JK. Dental association of C-reactive protein with cardiovascular dis-
infection and the risk of new coronary events: prospective ease risk factors and subclinical disease in healthy elderly
study of patients with documented coronary artery disease. subjects. Arterioscler Thromb Vasc Biol 1997: 17: 2167
Clin Infect Dis 1995: 20: 588592. 2176.
24. Mattila KJ. Viral and bacterial infections in patients with 34. World Health Organization. The world health report 1995:
acute myocardia infarction. J Intern Med 1989: 225: 293296. bridging the gaps. Geneva: WHO, 1995: 1.

120