Relationship between the burden of major periodontal bacteria and serum lipid profile in a cross-sectional Japanese study

Recent studies have revealed that chronic marginal periodontitis is associated with atherosclerosis (as an intermediate endpoint) and cardiovascular diseases [1‐3]. Periodontitis and its severity have been operationally defined by two types of clinical measurements, i.e., pocket depth and/or clinical attachment loss, in population studies [4, 5]. The clinical definition of periodontitis is dependent on the patient’s condition or the characteristics of the target study population, without any established gold standard. Thus, more valid, more reliable, and various measures for the chronic periodontitis associated with cardiovascular diseases seems to be required. Previous studies have suggested that the serum antibody levels against specific periodontal bacteria, such as Porphyromonas gingivalis and Aggregatibacter actinomycetemcomitans, or periodontitis-related microorganism groups [6] are useful to predict the risk of atherosclerotic diseases [7‐9]. Furthermore, periodontal bacteria harbored in gingival pockets, in regard to the burden of total bacterial species possessed by an individual, has been suggested to be a risk factor for atherosclerosis [10] and myocardial infarction [1].

As for major periodontal pathogenic bacteria, P. gingivalis, Prevotella intermedia, and Treponema denticola, as well as several other species, have been suggested to be possible causative agents in an in vivo study of atherosclerosis [11]. Another recent study [12] showed that specific bacterial groups, including the Orange-Red complex of P. intermedia, P. gingivalis, T. denticola, and Tannerella forsythia, were associated with elevated plasma glucose levels in adults. However, it did not show that those bacterial groups were related to other metabolic syndrome components. In contrast, another recent study [13] suggested that periodontitis, as determined by pocket depth, clinical insertion level, and bleeding on probing, was associated with distorted serum lipid levels, and also noted that the serum levels of P. gingivalis and A. actinomycetemcomitans antibodies may be a risk factor for decrease high-density lipoprotein (HDL) cholesterol levels. In light of metabolic pathways, P. gingivalis and its vesicles, for example, promote multiple cytokines as chronic infection and inflammation status resulting in elevation of triglyceride rich lipoprotein affecting on other lipoproteins. In turn, increase of triglyceride raises low-density lipoprotein (LDL) binding to macrophages and induce macrophages to modify native LDL, which plays an important role in foam cell formation [14, 15]. Notwithstanding, the association of periodontal pathogenic bacteria with serum lipid profile alteration as an atherosclerotic risk factor has not been well investigated, even though periodontitis is thought to be tightly related to the development of atherosclerosis. In addition, the effects of periodontal destruction, as a clinical parameter of the relationship between exposure to periodontal bacteria and lipid profiles in humans, are not well known. Therefore, in the present study, we hypothesized that exposure to periodontal bacteria could alter serum lipid metabolic pathways. We aimed to investigate the relationship between the prevalence of periodontal bacterial species with the total periodontal bacterial burden and the serum lipid profile, and to assess the combined effects of bacterial factors and the presence of periodontitis on the level of serum lipids.

Routine blood tests were used to measure levels of HDL, triglycerides (TG), and LDL cholesterol in serum, with total cholesterol (TC) levels calculated using the following formula: HDL + LDL + TG/5. These 4 variables (HDL, TG, LDL, and TC) were used as outcomes. Fasting glucose level was also assessed to define diabetes (fasting glucose ≥126 mg/L, or use of diabetic medication). Body mass index (BMI) was calculated by measurement of height and body weight [23]. Well-trained physicians measured systolic (SBP) and diastolic (DBP) blood pressure. Hypertension was defined as SBP ≥ 140 mmHg and/or DBP ≥ 90 mmHg or use of antihypertensive treatment [24]. In addition, lifestyle details, such as smoking, drinking, and regular exercise, were obtained using a standardized questionnaire by trained physicians or nurses in face-to-face interviews, as previously described [24]. Smoking and drinking were further classified into current, former, and never. The number of times of exercising per week was categorized as none, once or twice, and more than twice.

In terms of the operational definition of bacterial burden, the total bacterial burden of periodontal pathogenicity in this study was defined in cases in which the 4 types of periodontal bacteria were found in the saliva. A person was classified as having a high bacterial burden when all 4 bacteria were found in his/her saliva, which indicates a relatively higher exposure to periodontal pathogens. Therefore, the presence of a bacterial species scored 1 point for any of the 4 periodontal bacteria; exposure was considered as low (none of the 4 bacterial species present), moderate (presence of 1–3 bacterial species), or high (all bacterial species present) for statistical analyses.

In order to measure the statistically independent strengths of the association of bacterial burden on lipids, general linear models (GLM) were constructed after adjustment for potential confounding variables, including age, sex, periodontitis, BMI, diabetes, hypertension, smoking, drinking, exercise, tooth brushing, and dental flossing. To observe the change in effect size, confounders were added step by step into the models and the least square mean (LSM) values for lipid levels were also employed. Their p for trend values were tested as well. Lastly, LSMs were calculated again for the 6 groups categorized by bacterial strains and periodontitis after adjustment for age, sex, BMI, diabetes, hypertension, smoking, drinking, exercise, tooth brushing, and dental flossing for the combined effect on lipid level. The p for trend values were also determined.

Regarding of GLM analysis, the distribution of lipids variables was initially checked whether or not they were normal. The lipids were not normally distributed as a matter of fact so that they were transformed by square root, inverse square root, and natural log. The transformed results were compared to those of GLM. For preference of robust model, GLM analysis was finally selected. Normality test for outcomes, Shapiro-Wilk, Kolmogorov-Smirnov, Cramer-von Mises and Anderson-Darling tests were used for each 4 lipid type. And then the outcomes were transformed by square root, inverse square root, and natural log. The findings of GLM analysis in transformed data were compared to those in unconverted data (not shown).

As shown in Table 2, we examined 385 Japanese study participants (176 men and 209 women, average age 69.2 years). Those with greater numbers of different species of periodontal bacteria were more likely to be a current or ex-smoker (p = 0.05) and had periodontitis (p < 0.01). Periodontitis group showed a tendency to possess all of the periodontal bacteria, while P. intermedia was more likely to be harbored by ex-smokers and a higher percentage of older people possessed P. gingivalis.

Table 3 shows the crude associations between the periodontal bacteria and lipid levels. There were significant relationships between the average HDL level and the presence of each bacterium, except for P. gingivalis, the presence of 4 bacteria, and the co-existence of bacteria and clinical periodontitis. Furthermore, the concentration of TG was shown to be significantly elevated along with the increase in the bacterial burden.

In multivariable analysis, the presence of all 4 bacterial species was negatively associated with HDL, after adjustment for clinically defined periodontitis and other potential confounding factors, as shown in Table 4 and Fig. 2. Adjusted mean HDL levels according to the bacterial burn group (low, moderate, and high) were significantly reduced (p for trend = 0.05) (Table 4). Co-existing periodontal bacteria and clinical periodontitis (periodontitis [−] and periodontitis [+]) was significantly related to decreased HDL (Fig. 2). The adjusted mean HDL levels in patients with a combination of periodontitis (−) and bacterial burden (low, moderate, and high) and periodontitis (+) with the presence of bacterial species (low, moderate, and high) were decreased (p for trend < 0.5), while TG levels in those groups showed a trend to increase (p for trend < 0.5) as shown in Fig. 3. In reference to transformed lipids analysis, the final results were similar to those shown in Figs. 2 and 3, even though the estimated mean values were slightly different (not shown).

When HDL, as a main outcome, was used for calculation with the group means of bacterial burden, the power of this study exceeded 0.80 as the sample size ranged from 212 to 556 in two-sided tests (not shown).

Periodontal health and disease status have not been viewed in the context of a single periodontal pathogen, such as P. gingivalis, but rather in terms of the total pathogenicity of a biofilm [25]. The accumulated total burden of the harbored bacterial species has been suggested to complicate the bacterial community and is involved in the microbial shift from symbiosis to dysbiosis [26]. In the present study, persons who harbored all of the 4 different species showed poor HDL values after adjustment for clinically defined periodontitis and other potential behavioral factors. Additionally, the combination of harboring periodontal bacteria and periodontitis on decreased HDL and increased TG showed significant trends. Our results provide important epidemiological evidence that the accumulated burden of periodontal bacteria from saliva can independently affect regulation of HDL and TG levels, which are critical risk factors for atherosclerotic diseases, regardless of periodontitis. Such findings have not been reported previously, even though it has been shown that the presence of periodontitis, as defined by typical clinical measures, can contribute to a decrease in serum HDL and increase in TG [14, 27, 28], as well as to the development of atherosclerosis-related cardiovascular disease [29].

Interest in the role of the total periodontal bacterial burden, rather than that of a specific pathogen, in development of cardiovascular disease has emerged from studies reporting only one or two periodontal bacteria in relation to the occurrence of atherogenesis [30] and coronary heart disease [31]. Several recent case−control studies [32‐35] and a cohort study [1] have investigated the role of co-existing major periodontal bacteria in a cardiovascular disease event, and their findings have suggested that a higher inflammatory burden of periodontal bacteria could elevate the risk of stroke, myocardial infarction, and coronary artery disease.

In addition to the importance of the impact of bacterial exposure on the development of cardiovascular diseases as an end-point health outcome, it is vital to highlight biomarkers associated with atherosclerosis medication, such as serum lipid profiles, to gain a better understanding on how infection and/or inflammation induced by periodontal bacteria play a role in altering lipid metabolism. Previous reports have noted that patients with periodontitis diagnosed based on clinical parameters have a risk of dyslipidemia, such as decreased HDL, increased TG, and increased LDL [27, 29, 36]. In addition, control of periodontal inflammation by standard treatment was shown to improve HDL and LDL levels [37], and findings of a recent meta-analysis [38] indicated that periodontal treatment enhanced the levels of atherosclerotic biomarkers, including HDL, TG, and TC.

The present findings provide evidence suggesting that the total inflammation burden induced by exposure to major periodontal bacteria could induce dyslipidemia by lowering HDL and possibly increasing TG, after adjustment for periodontitis as a clinical parameter. The mean differences in HDL lowered by bacterial burden seem to be clinically small; thus, the effect of the oral pathogenic burden on low HDL or hyper-triglycerides beyond the normal range is unclear. However, we could not ignore this independent, albeit small effect from exposure to oral infection or inflammation, which may trigger the development of dyslipidemia and even atherosclerosis in individuals harboring several risk factors [14, 29]. A recent cross-sectional study [13] reported that the presence of A. actinomycetemcomitans and P. gingivalis reduced the plasma level of HDL, while another study [39], which used a scoring method for 3 pathogen antibodies, combined with herpes simplex virus, showed association with decreased HDL. In a Japanese prospective study [40], increased tooth brushing frequency caused improvement in hypertriglyceridemia. Consistent with the findings from previous studies, our findings established that inflammation related to oral bacteria contributes to deterioration of the lipid metabolic pathway, although some evidence regarding this association remains lacking. Nevertheless, an important strength of this study is empirical evidence showing a connection between the total burden of exposure to P. intermedia, P. gingivalis, T. denticola, and T. forsythia, detected from stimulated saliva, with decreased HDL and a trend for increasing TG in serum. Saliva seems to be the sample most readily obtainable from a number of subjects, and the bacterial burden, which can deflect the metabolic pathway, must be sufficient for bacterial detection in saliva (even though plaque specimens are a better medium for sensitive bacterial detection) [18, 24].

In this study, we exploited GLM models, with calculation of LSMs, although the level of lipids was not normally distributed, as we were concerned about the major limitation of transformed data (lack of a real value, which makes it difficult to interpret). We assume that the 4 lipid values were normally distributed, because the distribution of the lipids in the general population is normal and, furthermore, the implication of such a robust model is easier to understand.

This study had some limitations. Since it was cross-sectional in nature, we were not able to examine the causal relationship between exposure and outcome, and the interplay between periodontal inflammation and lipid dysregulation via host response reaction is complex. To improve our understanding of the role of the oral bacterial burden in lipid metabolism, well-designed studies, such as prospective or randomized controlled trials, are needed to verify these results. On the other hand, the sample size of this study was small, as the sample size calculation was implemented after the study had been finished, rather than prior to commencement, as this was an exploratory study. Based on the results of this study, statistical power was confirmed using final sample size whether the power was at least 0.80 or not. In addition, this study focused on only 4 primary bacterial species related to periodontitis, originating from gingival pockets. Nevertheless, other bacterial species have been found to be associated with the risk of systemic disease [12]; for this reason, pathogenic burden levels were classified as low, moderate, or high, instead of 0, 1–3, and 4. Despite this disadvantage, the present results suggest that a clinical bacterial test for the 4 well-known periodontal bacteria could be useful for screening patients with a greater risk of serum lipid profile deterioration and development of atherosclerotic disease. It is possible that oral microorganisms interact with each other; thus, each organism should be quantified to assess the individual contribution to the total pathogenic burden and modification of lipid metabolism. Moreover, no information was obtained about changed host responses, such as inflammatory serum cytokines, induced by periodontal infection and that resulted in lipid metabolic modification. Other variables may also not have been assessed, even though some crucial potential confounders were adjusted for in our statistical models. For instance, reduced chewing ability related to low number of remain teeth affects proper mastication, which also influences lipids metabolism negatively considering inclusion criteria (10 teeth and more in mouth). Finally, generalization of the study results may be limited, as only a small number of elderly Japanese persons living a metropolitan area were examined in this study. Further studies that consider host metabolic dysfunction after exposure to major periodontal pathogens, and that also include various populations, as well as the general population of Japan, with larger sample sizes, are necessary.

The results of this research suggest that a higher burden of bacteria was independently associated with lower HDL in serum after adjustment for periodontitis and other potential risk factors. This could show a possible link between exposure to major periodontal bacteria, in terms of the total burden, and increased risk of dysregulation of serum lipid metabolism. Furthermore, screening for the presence of periodontal bacteria could be beneficial to detect lipid changes earlier than atherogenesis, thus reducing the risk for development of atherosclerotic disease. Future studies that provide additional evidence about these issues are needed.