OUP user menu

European workshop in periodontal health and cardiovascular disease consensus document

Phillipe Bouchard, Pierre Boutouyrie, Francesco D'Aiuto, John Deanfield, Efthymios Deliargyris, Francisco Fernandez-Avilés, Francis Hughes, Phoebus Madianos, Stefan Renvert, Mariano Sanz
DOI: http://dx.doi.org/10.1093/eurheartj/suq001 B13-B22 First published online: 7 April 2010

Abstract

There is evidence from epidemiological research on the association between periodontal diseases (PD) and cardiovascular disease (CVD). In spite of these significant associations, however, there is still a lack of awareness in the cardiovascular community on their possible importance. In view of this evidence, an expert panel composed by six European periodontists and four cardiologists addressed the following questions: a) How important is the association of PD and CVD?; b) How do we measure exposure (PD) and outcome (CVD)?; c) What is the pathogenic link?; d) What is the relative importance of periodontal therapy as prevention of CVD?; e) Is it justified to recommend periodontal health with the goal of reducing CV risk?; f) What clinical and experimental research is needed?.

The general conclusions were that even though these proven epidemiological association between PD and CVD, there is, however, no compelling evidence that preventive periodontal care or therapeutic intervention will influence cardiac health. As Periodontitis continues to have a high prevalence within the population and the fact that CVD remains as the major cause of human death in developed countries, in light of these associations we can legitimately, based on evidence, state that oral health has an influence on systemic health in general and in CVD in particular, and therefore, we should promote oral health in general and periodontal health in particular as part of a healthy life style and hence as an important component in the prevention of CVD.

  • Periodontal diseases
  • Cardiovascular disease
  • Oral health
  • Systemic inflammation
  • Cardiovascular events
  • Periodontal pathogens

Introduction

Most of the epidemiological research on the association between periodontal diseases (PD) and cardiovascular disease (CVD) has been carried out in case–control and cross-sectional studies and the majority, although not all studies, have demonstrated significant associations between the two diseases. In fact, several meta-analyses showed that these associations are statistically significant and consistent, although of low magnitude. An important reason is the lack of a uniform definition of periodontitis, because different PD definitions are used in the studies leading to a wide range of exposure and hence, diluting the results. Nevertheless, because of the high prevalence of both PD and CVD, this association, even small, may be of great importance regarding its public health implications.

On one hand, there is still a need for better-designed prospective cohort studies; however, these studies need a long time (6–10 years) to get events and disproportionate effort to confirm an association, which has been shown to be significant, although low in intensity. The association is there and it is consistent, so there is maybe no compelling needs to devote time and resources in new large epidemiological studies. What is really compelling is to understand the pathophysiological link between PD and CVD, together with the possibilities to influence CVD through periodontal care.

Although there is a clear biological association between chronic inflammation and CVD on one hand and between PD and chronic inflammation on the other, it will be difficult to demonstrate the association between CVD in general, atherosclerosis in particular and PD, in terms of clinical outcomes. To start discerning the biology and mechanisms underlying the association of periodontitis with atherosclerosis, it is necessary to understand the complexity of this latter disease, since many different factors (lipids, hypertension, diabetes, smoking, and so on) contribute to atheroma formation on a lifespan basis. All these risk factors may have a common pathway that includes endothelial dysfunction and local inflammation. Similarly, there are many things that are still unknown about PD, including what controls susceptibility. Although the subgingival biofilm causes a local inflammatory response, in most people this process will not result in progressive periodontitis and there are people in which periodontitis progresses more rapidly than others. There is emerging evidence that the regulation of the inflammatory response and the nature of the inflammation may be major determinants of susceptibility and hence of progression of periodontitis.

There are several common risk factors for CVD and PD, which may act equally on both diseases providing a pathophysiological link, without necessarily having a direct causal link, or there is a common phenotype that underlies both conditions. Furthermore, there might be some genetic predisposition to suffer from both diseases. Schaefer et al.1 recently provided evidence of the shared association of the CHD high-risk locus with aggressive periodontitis. Nevertheless, the possible common predisposition underlying both diseases does not negate a direct influence of PD on CVD as well.

A randomized controlled trial (RCT) with an intervention is probably the only way to identify the causal pathways. To design such study, we should identify the best stage of the PD to intervene, either in early to moderate or severe PD and to choose the most appropriate CV endpoint. The most convincing endpoint would be hard cardiovascular events, such as death, myocardial infarction, stroke, or need for a re-vascularization procedure. However, the large sample sizes and the long follow-up necessary when testing for differences in clinical endpoints present a major challenge. Identifying appropriate surrogate endpoints represents a more pragmatic approach. Surrogates assessing CV early stages in the evolution of atherosclerosis include the measurement of the carotid intimae media thickness (IMT), arterial stiffness, or the assessment of endothelial function through measurement of flow-mediated dilation of the brachial artery.

There is, however, a reasonable doubt on how something that takes a very long time to build up, such as the increase in carotid IMT, may reverse just by an oral mechanical intervention. Intervention must take time before any measurable effect of surrogate endpoints is observed. Although CV events might all be explained by inflammation, the possible direct contribution of the pathogens should also be taken into consideration. Several studies have localized pathogenic oral bacteria in atheroma, although such observations do not prove a causative role. Both direct infection and inflammation activating the host response could be possible mechanisms and the fact that PD is a chronic condition, fits in with the time life exposure to risk factors required to develop CVD.

In spite of these significant associations, there is still a lack of awareness in the cardiovascular community on their possible importance. If the cardiologist could see the infection and destruction occurring in the dental pocket as a result of PD, they might understand the different mechanisms linking PD to the biology of atherosclerosis.

In view of this evidence, the expert panel addressed the following questions:

  • How important is the association of PD and CVD?

  • How do we measure exposure (PD) and outcome (CVD)?

  • What is the pathogenic link?

  • What is the relative importance of periodontal therapy as prevention of CVD?

  • Is it justified to recommend periodontal health with the goal of reducing CV risk?

  • What clinical and experimental research is needed?

Epidemiology and association: how important is the association of periodontal disease and cardiovascular disease?

There are three systematic reviews recently published24 evaluating the evidence on these associations. The review by Bahekar et al.2 stratified the studies by design type and used either just prospective studies or case–control or cross-sectional studies. The weakness of the association can be appreciated because there are only five prospective studies, which included around 90 000 patients and the hazard ratio was 1.14 (i.e. the individuals with PD have a 1.14 times higher risk of developing CHD than the controls). In case–control studies, the risk of developing CHD rose to OR 2.2 and in cross-sectional studies the OR were in the range of 1.6. Prospective studies are inherently better to address the question of association. However, the existing prospective studies add less quality in terms of their design than case–control and cross-sectional studies, because they were not originally designed to address this association and the quality of PD assessment may be questionable. In fact, in this meta-analysis none of the studies had been designed to evaluate the relationship between PD and CVD and in many of these patients, the evaluation of exposure (PD) was self-reported. Moreover, all the OR mentioned, even if the association is significant, are small compared with classical surrogate endpoints like artery stiffness, which has an OR in the range of 2–3 depending on the population.5

These OR are adjusted for other cardiovascular risk factors; however, in an intervention RCT, crude OR, before adjustment, could be more interesting because we could consider PD as an integrator of CV risk, as it was done for arterial stiffness, IMT, or endothelial function.

All reviews and papers consistently come out with a positive statistical association in spite of the use of different PD definitions. Whenever PD was better assessed, the risk was greater and similarly when the outcomes were narrowed down, e.g. stroke rather than CVD. It is clear that we have enough case–control and cross-sectional studies, although we might need prospective studies with adequate assessment of exposure and outcomes. The problem is the time needed for these longitudinal studies to obtain events (6–10 years) and it might not be worthy in terms of resources and time, hence we should rather focus on intervention studies and the use of surrogates.

The use of surrogate endpoints may be still exploratory (hypotheses generating) but their use may be needed before designing clinical outcome-based trials. These latter trials need large amounts of time and resources and cannot be undertaken without previous knowledge on what to expect. The intervention study is even more justified, because the periodontal intervention will enhance oral health, which will improve general health by itself.

As a summary, we have agreed that the association between PD and CVD, although low, has consistently been shown significant in all systematic reviews published. Furthermore, the association is found in spite of the different measurements of exposure and outcomes used, which strengthens the apparently weak association. The association, even if it is low, is very important for the potential implication it has on public health due to the large prevalence of both conditions. There are common risk factors and common confounders, which make dissecting the specific association difficult and which we need to study to fully understand it.

How do we measure exposure (periodontal disease) and outcome (cardiovascular disease)?

Exposure

Periodontal disease is defined by the anatomical destruction of the tissues supporting the tooth that occurs following the disease process. Using this definition, severity is measured according to the amount of destruction. The final outcome is tooth loss and the intermediate outcome is the amount of periodontal ligament and/or bone loss that occurs. The difficulty in using tissue destruction as a definition of PD is that it does not take into account the disease process, but rather the consequences. As we are dealing with a chronic infection leading to systemic inflammation, we could alternatively define the disease by the infection load or by the systemic inflammatory consequences, thus dealing more with the process than with the secondary destruction.

Thus, the first issue would be the selection of the most appropriate way to measure PD, either by destruction (the disease's consequence) or by the exposure itself in terms of infection load or host response. There is still no consistent biomarker associated with the presence of disease that could measure exposure. The amount of destruction in itself correlates with the increase of the systemic manifestations of the infection. In gingivitis, the inflammatory lesion is confined to superficial tissue with no deep destruction and very little ulcerated epithelium. However, in PD and in presence of tissue destruction and an ulcerated periodontal pocket, there is a clear association with a whole range of biomarkers, such as pro-inflammatory cytokines that increase in quantity and can be measured locally or systemically.

A specific marker of systemic inflammation is not good enough because although it would be associated with the presence of disease, it would not give an idea of the history of the disease. Bacterial load, antibodies, C-reactive protein, and so on are all snapshots; whereas attachment on bone loss are cumulative measurements of what has happened over the years and it would have value in terms of a chronic disease like atherosclerosis, where some history is needed. A combined marker, something accounting for the duration of the disease process and at the same time incorporating the disease load (in terms of infection or inflammation) would be optimal, as two people could have the same level of PD but one be more systemically affected than the other.

Among the surrogates, there are specific markers like Porphyromonas gingivalis, which can only be traced in the mouth, but others, such as C-reactive protein or interleukin-1 (IL-1) that can be measured systemically, but are very non-specific since many conditions may elevate their levels. Porphyromonas gingivalis may be found in patients with minor periodontal problems and in that respect it might not be a good marker of disease. Its presence, however, is a marker of infection in the sense that a certain environment is required for its growth. Moreover, measuring bacteria in the mouth, although specific, is largely affected by technology. There are different ways to measure bacteria, all being difficult to standardize. Probably, the most accurate is nested-PCR but the DNA amplification is so large that even when it is positive, the significance is unclear. It is, therefore, advisable to utilize microbial diagnostic techniques that not only assess presence or absence of bacterial, but rather the load of specific pathogens.

In regards to the systemic effect of the infection burden, the oral bacteria load may not correlate the systemic effect, because the quantity of bacteria in the mouth may not be proportional to the systemic exposure to these bacteria. We need some reliable method to measure the presence of specific periodontal bacteria or bacterial products in serum.

We have a consensus on the need to evaluate exposure with a of combined measurement of PD severity (structure) and disease process through biological markers, probably more related to systemic inflammation than to infection, but we should not discard in the future a reliable measurement of systemic infection exposure. It does not have to be extremely specific, as long as it is measurable, reproducible and that can be used to track the oral disease activity. There are some salivary markers that look promising and if they correlated the amount of destruction to the amount of systemic inflammation, they would be very useful biomarkers.68

Outcome: cardiovascular disease surrogates

Death or other clinical events are the best endpoints in long-term cardiology studies but it takes time to get clinical events and thousands of patients. The factors leading to precipitation of the event (thrombosis) may be different from factors leading to atherosclerosis. There are surrogates, such as endothelial function, IMT, arterial stiffness, and so on, which are intermediate phenotypes or measures more closely related to outcomes than to the risk factors that may drive them. Cardiologists share similar problems than Periodontists, since they too are interested in structure, accumulated burden of disease and propensity-risk for events, all of which necessary for the understanding of the biology of atherosclerosis.

There are two distinct questions regarding the link between PD and CVD. First, is PD exposure related with progression of atherosclerosis and second is PD related to activation of atherosclerotic plaques and CV events? In order to assess disease progression, some measure (or surrogate) of structural disease is needed, such as endothelial function or arterial stiffness (pulse wave velocity, arterial waveform analysis) that are measures linked to the development of atherosclerosis and its progression.

Intimae media thickness is a useful measure of structural disease and accumulative burden disease, being an accepted marker of early atherosclerosis. It is an easy and non-invasive diagnostic technique and it has been significantly linked to stroke and CV outcomes. It is, however, not a good measure of the risk for plaque rupture, the underlying cause of most CV events. This risk is related to plaque composition, which can be assessed only invasively for the coronary arteries with the closest next surrogate for future events being the systemic inflammatory markers such as C-reactive protein. Regarding non-invasive imaging, molecular resonance imaging can be utilized for plaque composition visualization, but due to limited spatial resolution can only be used in large calibre arteries such as the carotids. In addition, the cost and time is prohibitive for large cohorts.

Arterial stiffness has been validated as an intermediate endpoint in more than 13 population trials.9 It integrates the ill effects of CV risk factors on large arteries along time. However, the number of intervention trials based on arterial stiffness measures is still limited.

Endothelial function has also been used as a surrogate marker and has demonstrated significant association with CV events.

Biological markers of systemic inflammation are also reliably linked to CV events. C-reactive protein is a widely accepted marker of inflammation and has been proposed as a surrogate marker for CVD.10 The problem with C-reactive protein is that it is associated with PD and it is, therefore, going to be already elevated. In CVD, C-reactive protein operates in populations to help additional risk stratification over a very small range of very low C-reactive protein. However, in PD, C-reactive protein is way above those levels. If the intervention (periodontal therapy) convincingly lowers C-reactive protein, it will be a direct consequence of oral benefits but not a measure of CV consequences. It is however, logical that when intervention trials are planned, C-reactive protein measurements should also be included.

As a summary, there is a need to clearly understand the relationship between periodontitis and progression of atherosclerosis. We may use several surrogates to aim to understand what drives to the cardiac event, but there are no predictive measures yet. A conjunction of techniques are necessary to characterize the surrogate endpoints, as no technique alone can answer the question and some are difficult to perform, like measuring endothelial function by means of brachial artery vasodilatation.

What is the pathogenic link?

Considering the relative importance of direct infection vs. systemic inflammation as the etiopathogenic link with atherosclerosis, there is consensus that we need to focus on systemic inflammation rather than on direct bacteria invasion/translocation through the systemic route. However, these two routes are interlinked, because there would be no systemic inflammation if there were no systemic bacteria exposure.

The epithelial ulcerations at the biofilm–gingival interface allows for oral bacteria to enter into the system and several studies1116 have demonstrated how these specific bacteria or bacterial DNA are deposited on the atheroma plaques, although we do not know the significance of this finding. The problem is that most papers looking for bacteria in atherosclerotic plaques have looked specifically for periodontal bacteria but not for any other and it may well be that all bacteria which get into the body, get deposited into the plaques.

Concerning whether infection could be the trigger for CVD, the evidence from epidemiology studies shows that infections increase the risk of CVD, irrespectively of the type of infection (e.g. respiratory or urinary tract infections). Thus, it seems that it is not the type of bacteria but the type of host response what matters; however, some studies have shown an association of CHD with infection by certain microorganisms but not with others.17

The intervention study by Tonetti et al.18 highlights the importance of bacteraemia, since when dental plaque was eliminated there was a transient acute systemic inflammatory response, suggesting the systemic reaction to the bacteraemia. The pathogens and the inflammatory pathway are not therefore, two separate things, in fact, in Periodontitis, it is the infection that causes inflammation.

The inflammatory hypothesis is well established in the aetiology of atherosclerosis. Non-infection inflammatory diseases, such as rheumatoid arthritis have also been associated with CVD.19 It is important to consider, however, that no intervention trials with systemic anti-inflammatory therapy (including steroids) have ever demonstrated clinical benefit. Furthermore, non-inflammatory mediators, such as oxidized LDL, might also trigger local endogenous inflammation within the vessel wall and local inflammation releases small quantities of short-lived mediators that may access the circulation, but probably would soon get diluted in the blood.

It is therefore, unlikely that these small quantities of mediators (IL-1 or TNF-α) have a significant contribution towards a systemic effect. If there are no systemic bacteraemia or bacteria products, it is unlikely that the local production of IL-1 or TNF-α in gingival tissue would have a significant effect in the rest of the body. Recently, scientific reports have been made on the effect of anti TNF-α therapy and regression of arterial stiffness and these results could validate the rationale for PD intervention.20

As a summary, the systemic inflammation pathway is today the most accepted and studied mechanism that could link the etiopathology of atheroma formation to the chronic inflammation derived from PD. The issue is whether the local inflammatory state of periodontal tissues is enough to cause systemic inflammation or whether we also need the systemic invasion of bacteria or bacterial products.

There is a third possible pathway, which is the modification of known cardiovascular risk factors. This pathway has been demonstrated in the metabolic syndrome21 where lipid profiles are modified and thus become dysfunctional and pro-atherogenic.22 It would be interesting to study whether the cardiovascular risk-factor profile of PD patients gets modified.

Whether one or the other pathways are different leads to distinct consequences. If bacteria or their products cause systemic inflammation leading to athero-genesis, the preventive or therapeutic strategy should be anti-bacterial. If, however, it is the local tissue inflammation the responsible of the systemic effect, the strategy should be anti-inflammatory (e.g. anti-metalloproteinases).

Is there evidence that periodontal therapy or prevention has a long-term benefit on cardiovascular health, both in terms of reduction of cardiovascular events and improvement in surrogate markers?

A recently reported interventional trial18 showed for the first time with an RCT that a modification of an oral event altered a CV marker like endothelial dysfunction. Thus, we can say that epidemiologically, endothelial dysfunction is associated with PD, and that treating PD in a particular population can alter the biology of the arterial wall, although we cannot say that periodontal therapy leads to long-term benefit in CV health.

The improvement of endothelial function was also corroborated by the significant decrease in E-selectin levels in the treatment group when compared with the controls. E-selectin is an adhesion molecule expressed in cytokine-activated endothelial cells, being a marker of endothelial activation. Also in this trial, flow-mediated dilatation (FMD) and tissue plasminogen activator/plaminogen-activator inhibitor type 1 ratio, changed in the same direction, reflecting also improvements in the endothelial adhesion and coagulation functions. The phenotype of the endothelium and all its functions, changed in the same direction, but the study does not provide information of the possible mechanisms explaining this outcome.

Another intervention trial that has raised controversy is the PAVE study conducted in the USA,23 since contrary to Tonetti's report18 it has shown a limited systemic positive response after the periodontal intervention. Two factors, the population selected and the efficacy of the periodontal intervention can explain the limited differences between the intensive-treatment group and control-treatment group shown in this study. The population were subjects in secondary prevention of CV events with little severity in PD and the intervention rendered was not very efficient, as the changes in the periodontal parameters after treatment were not very pronounced in both the treatment and the control groups.

In contrast, in Tonetti's study,18 the selected population had severe periodontitis and they used a combination of mechanical and antibiotic therapy demonstrating a highly significant difference in the periodontal outcomes after therapy between the experimental and control groups. The need of systemic antibiotics is, however, controversial, since in another intervention trial24 where periodontal therapy was purely mechanical, without antibiotics, it was also demonstrated a significant effect on endothelial function. If the benefit of the intervention is a reduction of the inflammation, long-term follow-up (at least 6 months) is needed and it is very important to select the right population.

Another surrogate marker for atherosclerosis used in intervention studies has been the measurement of the decrease in IMT in the carotid artery after periodontal intervention. Increased IMT is reversible; it can get thinner with cholesterol-lowering therapy2527 and the technique to measure it, image analysis, is now quite settled, making it a feasible examination to do in a trial. The study by Piconi et al.28 showed a decrease of IMT by over 10%, whereas Hodis et al.29 showed that for each 30 μm per year in IMT increase, the risk of CV events increased by two-fold in patients with coronary heart disease.

Another question is whether people with high risk of CV events (i.e. with hypertension, family history, and so on) may also have a risk for advanced periodontitis. We know that incidence of PD in obese population is much higher than in normal weight population and some studies link obesity with PD.30,31 In a recent publication,32 it was found that each individual component of the metabolic syndrome showed crude OR associated with diagnosis of moderate or severe PD. In this study, when models were adjusted with metabolic syndrome as the definition of diagnosis, the association was carried through. The only factor that kept the statistical significant association was glucose levels.

The reduction of CV events with statins is about 30%, but it varies on the initial levels of cholesterol or depending on whether it is primary or secondary prevention, etc. If you lower LDL cholesterol earlier you get more CV benefits. People with a genetic variable that lowers LDL-cholesterol level at about 20% from childhood (which is less than the reduction obtained by statins) showed an 88% reduction in CV events because it was a lifetime reduction in LDL cholesterol.33 In cardiology, statins are given to 50- to 70-year-old subjects and although there is a 25% risk decrease, still there is a 75% chance of events. The longer someone is on statins, the more the CV benefit. It is not only the degree to which LDL cholesterol is lowered but also the duration of this lowered LDL. In a similar way, if you manage PD over your lifetime, it will probably result in a greater benefit in terms of CVD outcomes.

As an additional remark, the effect of periodontal treatment on C-reactive protein from the meta-analysis of Ioannidou et al.34 was −0.18 mg/L (−0.70 to 0.35 95% CI), which is half of the reduction obtained at the JUPITER trial with the use of high doses of statins,10 so the mean effect of periodontal treatment is half of that of statins, which is not a small effect, especially with the limitation of the shorter time of type of intervention this is. We need more trials to update this meta-analysis and with further data we might see if that magnitude is greater or not, as this meta-analysis is based in only three studies.

In terms of periodontal therapy producing benefits, intensive (or effective) short-term periodontal therapy improved surrogate markers associated with CV risk. We do no know, however, what would be the long-term effect of periodontal intervention (the periodontal health maintenance-related outcome). This fact, however, would be difficult to demonstrate due to the unethical aspects of having untreated patients for long term. One way to go around that is to maintain the benefits of periodontal treatment and having those who quit treatment as the control group. Another way to overcome difficulties with the control group could be to take an intermediate step, similar to what was done in the classical periodontal treatment study carried out in Sweden35 where severe periodontal patients were selected and treated until they reached an expected outcome at which point they were either sent to their primary clinics or kept for their maintenance for 6 years (not randomized) and there was a great difference in periodontal outcomes. This was a non-randomized study, but it could be a model to look at surrogate markers in the long term starting from periodontal health rather than from oral disease.

As a summary, we have identified some markers related to structure in terms of vessels: endothelial function, IMT, and arterial stiffness, and some outcomes in terms of events; such as stroke that might be easier than specific events like death from myocardial infarction.

Is it justified recommending oral and periodontal health with the goal of reducing cardiovascular risk?

In the last 2–3 years, major oral health companies have used the link between PDs and CVD as their main marketing message stating that oral health promotes systemic health (more than CV health). In light of what we know today, we should question whether it is appropriate to make recommendations of preventive programs in oral and periodontal health for the sake of CV health. There is no compelling evidence that preventive dental care will influence cardiac health, in spite of the clear links between these two conditions. We still do not understand properly the real impact of oral health or periodontal intervention in terms of CV events, so it is not advisable to promote the message of oral health with the specific goal of preventing CV events, because it is too early to make that claim. One could say, however, that since oral infection and inflammation is not good for someone health in general, promoting oral health would be positive. In fact, having a healthy mouth is important for life. Some studies have shown that oral disease is associated with increased mortality3638 and there is evidence that PD affects quality of life,3941 with periodontal treatment being able to improve it.

Although it seems reasonable that when the cardiologist identifies in his clinic someone with severe PD could recommend him to seek therapy, it is too early to establish this recommendation as part of the routine heart examination. For patients with established CV disease, it is not unreasonable that the cardiologist raises his bar of awareness and puts PD on his checklist and could suggest to this patient that taking care of PD might also help.

Legitimately, based on evidence, we can say that: (i) oral inflammatory disease or PD has wider effects than previously appreciated; (ii) systemic inflammation results from that and it can cause adverse CV wall changes, and (iii) resolution of oral disease can result in reversal of those CV wall changes. We do not have evidence, however, to say that oral health may prevent future CV disease.

In summary, we have agreed that oral health has an influence on systemic health in general and in CVD in particular, and this is an important message in order to promote prevention more than to promote treatment. Thus, promoting oral health as part of a healthy life style is a sensible recommendation.

What is the clinical and experimental research needed to provide definitive evidence?

In terms of selecting, the appropriate study population, there are different target groups in both disease ends. In the periodontal end, we could select severe PD patients or a population with known susceptibility for the disease and once treated and periodontal healthy being followed long term. In the CVD end, we could select people who are healthy and do not have elevated classical risk factors, people who are healthy but have high risk factors and people who have had a CV event and are starting secondary cardiac prevention.

In the outcome selection, there is no single intermediate phenotype or surrogate measure that will provide definite evidence to the cardiology community, because there has not been a clear demonstration in terms of the reduction of cardiac events. So for definite evidence, the intervention will have to reduce clinical CV events. The secondary prevention group is the logical target group, since it is the one in which more events are expected, thus facilitating the evidence of a treatment benefit (by reduced number of events). Therefore, if an intervention study of PD is to be undertaken we should specifically select patients with acute coronary syndrome, who have an increased 6–12 months residual risk after the initial event.

The Interheart study42 showed that 94% of women and 92% of men who had heart attacks had one of eight modifiable environmental risk factors. One problem with a population at high CV risk is that they would be on concomitant treatments, which will dilute the effect of the oral intervention. This fact has to be taken into account in the sample size calculation.

Another important issue is patient selection (hospital, GP clinic, dental clinic), as it will modulate the population selected. A selection of CV high-risk patients who at the same time have also moderate–severe PD will be encountered more easily within hospital facilities. As the screening and patient selection is based on the diagnosis of PD, the correct strategy would be to have a recruiting dentist at the cardiology ward, screening rapidly all patients fulfilling inclusion and after that selecting them and providing periodontal care.

The three issues important to this trial are: (i) recruiting a population with sufficient future CV risk to have enough events, (problem being that this population will be on medication that will lower the CV event rate). (ii) Patients should also have enough PD on which the intervention can act. (iii) The periodontal treatment should provide sufficient chronic separation of the groups to prove a difference. As CV endpoints we could have any major CVD event: new angina, stroke, heart attack, death, a combination of events, and so on. Data from the TNT trial43 show that in people with chronic current coronary heart disease, the risk of dying form heart attack in 5 years is very low. In people with moderate risk, which are persons in whom treatment could be paradoxically more effective, event rate is low and often overestimated, and the risk of trial failure is high. In such groups, the typical incidence for primary cardiac event is 1% in 10 years. The group we would like to recruit should have a risk of 10% in 10 years to have at least 100 events in the lowest incidence group. We will therefore include stroke as one of the outcomes, which will increase events (composite CV events). We would suggest excluding soft endpoints such as new onset of angina or revascularization procedure (Coronary Artery Bypass Graft Surgery (CABG) or angioplasty).

The ethical question is also important. When the cardiologist tells someone who just had a heart attack that he needs to check his teeth or sets up a dentist appointment for them, the patient is offered access to care, which normally would not be sought for. This could be taken as an advantage of beneficiating of additional care, however, the control group is denied access to such care and this could pose also an ethical issue. As the ethical problem with the control (no treatment) comes from the selection criterion of having a certain amount of PD and from the obligation to fulfil a standard of periodontal care, one possible solution could be to give them all periodontal treatment and then randomize to institution (optimized) dental care or to their local (standard) dental care provider during 5 years.

Alternatively, similar to the PAVE study design,23 we could select people with high risk in secondary prevention, perform an examination to see the degree of disease by our defined threshold and decide two groups: one group would receive treatment in the hospital or university and the other would only be advised that they need treatment, which they might take from their regular practitioner. If you just tell the patient that they need periodontal treatment, probably only 10–15% would go to get it. This way would reduce the number of patients to include and would solve the ethical problem, because it is how it is done in clinical practice. Usually, the patient is referred from the GP to the specialist, and in the study, we would be telling patient in the control group to go to the GP, who should refer them to the specialist. You are not denying treatment. In addition, you cannot assure that the intensive treatment will have an effect on the patient's risk of CV events. We must consider that if to the aim is to demonstrate a prevention of CV events; we need a very long duration studies with large populations. The aforementioned studies on surrogate endpoints such as IMT, endothelial function, or arterial stiffness could be considered as preparatory to large-scale randomized clinical trial with hard clinical events as the endpoint.

When designing these trials, we are dealing with two models: in the first model, once patients are selected they would be randomized to treatment or sent to their oral health provider. In the second model, once patients are selected, they will be periodontally treated and once they get oral health they would be randomized to close maintenance care and the other sent to regular oral health provider. The second model needs more patients and longer time because the difference of the oral disease that will be experience over 5 years is limited making the first model more feasible.

Another study design would be to recruit patients with high CV risk, without previous oral examination, and randomize them to either having the periodontal assessment (and receive whatever treatment is considered appropriate) or not (and go home). The experimental group, the one who has received the assessment, will have two subgroups, the healthy and the diseased; this latter group will receive whatever treatment is needed and will be the test group. Randomizing the assessment instead of the treatment will need more patients and the range of disease will be less severe but it avoids all ethical issues. The number of needed patients will depend on the prevalence of PD in the population. The difficulty lies in that the number of patients in the non-assessed group will be greater than those who will receive treatment. As the treated group should be kept healthy all the time and in enough number to show a number of events as opposed to the non-assessed group, recruitment should be in a 2:1 ratio.

This study design has extra attraction in terms of public health because compared with the previous intervention studies taking people with PD, in this one, intense periodontal intervention will be carried on in patients who come through the cardiac ward to see if the intervention really works. It will also be much easier to recruit patients and not much has to be done with the non-assessed group, just to follow them up. Dental treatment should be simple, so that it does not have to be done in the specialist clinic (depending on the country, maybe a dental hygienist could do it), but it has to be the best possible treatment and even those with limited PD should be treated. Thus, the goal should be to provide the best therapy that can be achieved and each country could do it its way, but always providing this high-quality care.

Regarding timing of periodontal treatment, we have to take into consideration that most people who have had a CV event leave the hospital with stents on double anti-platelet therapy; thus, they may bleed profusely if their gums are probed. Currently, it is known that the more we give this kind of anti-platelet therapy, the better these patients do; thus, patients take it for 6 up to 12–18 months. With a drug-eluting stent, which is 60–70% of all stents that cardiologists implant in Europe, one needs at least 9 months of double anti-platelet effect, according to recommendation of all cardiac societies. The consensus by the American Heart Association, American College of Cardiology, Society for Cardiovascular Angiography and Interventions, American College of Surgeons, and the American Dental Society published in Circulation44 recommended not to stop the anti-platelet medication after placement of a drug-eluting stent and educating the patient and healthcare providers about hazards of premature discontinuation. In practice, not all dentists will accept to do a procedure in patients maintaining double anti-platelet medication.

Regarding the appropriate timing when it is safe to do periodontal therapy, it depends on whether the dentist accepts to do the procedure under the double anti-platelet effect. If he does, from the CV point of view it could be done almost immediately after discharge. These patients usually have a first follow-up with the cardiologist at 1 month after discharge from the hospital, as it is the highest event rate month and it is also the time when the person changes his life style and is taking different medications that might have side effects. If after this first follow-up at 4–6 weeks, things are stable, they are complying with therapy and had not had complications, dental intervention could then be undertaken.

The non-surgical periodontal therapy: should be performed with ultrasounds, where tissues are not traumatized as much and it is equally effective, as hand instruments. Perhaps a combination with some anti-microbial agent would help efficacy. The patient will need local anaesthesia and may need prophylactic antibiotics, as with the procedure we are going to induce bacteraemia in patients, who just had a cardiac problem. The treatment should be staged trying to minimize the C-reactive protein rise because producing acute inflammatory stimulus in an unstable patient could be a potential ethical issue.

General conclusions

Epidemiological studies have clearly shown a moderate but significant association between Periodontitis and CVD. There is, however, no compelling evidence that preventive periodontal care or therapeutic intervention will influence cardiac health and since we still do not understand properly the real impact of oral health or periodontal intervention in terms of CV events, it is not advisable to promote the message of oral health with the specific goal of preventing CV events.

As Periodontitis continues to have a high prevalence within the population and the fact that CVD remains as the major cause of human death in developed countries, in light of these associations we can legitimately, based on evidence, state that oral health has an influence on systemic health in general and in CVD in particular, and therefore we should promote oral health in general and periodontal health in particular as part of a healthy life style and hence as an important component in the prevention of CVD.

Conflict of interest: none declared.

Funding

This workshop has been made possible by an unrestricted grant from Colgate Palmolive Europe.

References

View Abstract