Literature search
The literature search identified 617 relevant publications. A total of 162 duplicates were removed. Screening the titles and abstracts resulted in the elimination of 267 studies that failed to meet any of the inclusion criteria, and all proceedings and books were removed. We also excluded all animal experiments and in vitro experiments. Fifteen reviews and meta-analyses were removed. A total of 101 papers were selected for full-text screening. In 17 articles, Mets or CHD was used as the research object but without a blood lipid index, and 15 articles did not include data on the relationship between periodontitis and blood lipids; we excluded these articles from our current investigation. Four studies contained repetitive populations, and we used the most recent study with the largest study population. After the quadratic search for reviews, we finally identified 67 articles. (Fig. 1).
Association Between Dyslipidemia And Periodontitis: Periodontitis As The Outcome
TC and periodontitis
Twenty-three studies evaluated the difference in the serum TC level between periodontitis and healthy control groups with the mean (SD). TC levels were higher in patients with periodontitis than in controls, with a pooled mean difference of 12.45 mg/dL (95%-CI: 7.55, 17.34, p < 0.01). There was significant heterogeneity between the studies (I2 = 88%, p < 0.01) (Fig. 2-a). Meta-regression showed that the sources of the heterogeneity may be diagnosis of periodontitis (P diagnosis) (p = 0.041) and BMI matching (p = 0.050) (Table S5). The result was robust regardless if any one study was omitted (Supplementary Fig. 1-a). No significant publication bias was found after Egger’s (p = 0.093) and Begg’s tests (p = 0.476).
Sixteen studies reported the OR values to evaluate the association between the serum TC level and periodontitis. The pooled OR was 1.28 (95%-CI: 1.09, 1.50, p < 0.01), and substantial heterogeneity between the studies was found (I2 = 71%, p < 0.01) (Fig. 3-a), indicating that a high TC level is a risk factor for periodontitis. Meta-regression showed that the sources of the heterogeneity were the year of publication (p < 0.001), study design (p < 0.001), age ratio (p < 0.001), P diagnosis (p = 0.001) and adjusted OR (p = 0.001) (Table S5). The result was robust regardless if any one study was omitted (Supplementary Fig. 2-a). A significant publication bias was found after Egger’s (p < 0.001) and Begg’s tests (p = 0.021).
Tg And Periodontitis
Twenty-four studies evaluated the difference in the serum TG level between periodontitis and healthy control groups with the mean (SD). TG levels were higher in periodontitis patients than in controls, with a pooled mean difference of 26.08 mg/dL (95%-CI: 17.23, 34.93, p < 0.01). There was significant heterogeneity between the studies (I2 = 100%, p = 0) (Fig. 2-b). The meta-regression did not find any significant source of heterogeneity (Table S5). The result was robust regardless if any one study was omitted (Supplementary Fig. 1-b). No significant publication bias was found after Egger’s(p = 0.366) and Begg’s tests (p = 0.197).
Seventeen studies reported the OR values to evaluate the association between the serum TG level and periodontitis. The pooled odds ratio was 1.65 (95%-CI: 1.36, 2.02). We used the random effect model due to the presence of heterogeneity between studies (I2 = 72%, p < 0.01) (Fig. 3-b), indicating that TG levels were significantly associated with periodontitis and that high TG levels are a risk factor for periodontitis. Meta-regression showed that the sources of the heterogeneity were the year of publication (p = 0.004), study design (p < 0.001), total sample size (p < 0.001), age (p = 0.001), P diagnosis (p = 0.001) and adjusted OR (p = 0.001) (Table S5). The result was robust regardless of if any one study was omitted (Supplementary Fig. 2-b). A significant publication bias was found after Egger’s(p = 0.003) and Begg’s tests(p = 0.006) .
Ldl And Periodontitis
Twenty studies evaluated the difference in the serum LDL level between periodontitis and healthy control groups with the mean (SD). LDL levels were higher in periodontitis patients than in controls, with a pooled mean difference of 7.35 mg/dL (95% CI: 2.70, 12.00, p < 0.01). There was significant heterogeneity between the studies (I2 = 93%, p < 0.01). (Fig. 2-c) However, the meta-regression did not find any significant source of heterogeneity (Table S5). The result was robust regardless if any one study was omitted (Supplementary Fig. 1-c). No significant publication bias was found after Egger’s(p = 0.208) and Begg’s tests (p = 0.650).
Nine studies reported the OR values to evaluate the association between the serum LDL level and periodontitis. The pooled OR was 2.62 (95%-CI: 1.44, 4.77, p < 0.01), indicating that LDL levels were significantly associated with periodontitis and that high LDL levels are a risk factor for periodontitis. We used the random effect model due to the presence of heterogeneity between studies (I2 = 69%, p < 0.01) (Fig. 3-c). Meta-regression showed that the sources of the heterogeneity were the study design (p < 0.001), total sample size (p = 0.014), and P diagnosis (p = 0.018) (Table S5). The result was robust regardless if any one study was omitted (Supplementary Fig. 2-c). A significant publication bias was found after Egger’s(p = 0.015) and Begg’s tests (p = 0.312).
Hdl And Periodontitis
Twenty-two studies evaluated the difference in the serum HDL level between periodontitis and healthy control groups with the mean (SD). The HDL levels were lower in periodontitis patients, and the pooled mean difference for the HDL levels in the periodontitis patients and healthy control groups was − 3.71 mg/dL (95%-CI: -4.83, -2.59 mg/dL, p < 0.01). There was significant heterogeneity between the studies (I2 = 64%, p < 0.01) (Fig. 2-d). Meta-regression showed that age may be the source of the heterogeneity (p = 0.003) (Table S5). The result was robust regardless if any one study was omitted (Supplementary Fig. 1-d). No significant publication bias was found after Egger’s(p = 0.670) and Begg’s tests (p = 0.844).
Thirteen studies reported the OR values to evaluate the association between the serum HDL level and periodontitis. The pooled OR was 1.24 (95% CI: 1.15, 1.35, p < 0.01), indicating that a low HDL level is a risk factor for periodontitis. There was no significant evidence for heterogeneity between the studies (I2 = 0%, p = 0.64) (Fig. 3-d). The result was robust regardless if any one study was omitted (Supplementary Fig. 2-d). A significant publication bias was found after Egger’s(p = 0.030) and Begg’s tests (p = 0.961).
Vldl And Periodontitis
Four studies evaluated the difference in the serum VLDL level between periodontitis and healthy control groups with the mean (SD). The VLDL levels were higher in periodontitis patients than in controls, with a pooled mean difference of 8.69 mg/dL (95% CI: 2.20, 15.18, p < 0.01). There was significant heterogeneity between the studies (I2 = 81%, p < 0.01) (Fig. 2-e). No significant publication bias was found after Egger’s(p = 0.266). and Begg’s tests (p = 0.174).
Some studies were not included in our meta-analysis due to the lack of information utilized. Saxlin T reported an association between high serum triglycerides and low HDL-cholesterol levels with periodontal pockets by quintiles (15). Akkaloori Anitha stated that the mean LDL and VLDL levels were significantly higher and the HDL levels were lower in periodontal patients than in healthy controls (16).
Association Between Dyslipidemia And Periodontitis: Dyslipidemia As The Outcome
Periodontitis and dyslipidemia
Three studies reported OR values to evaluate the association between periodontitis and dyslipidemia. Periodontitis was a risk factor for abnormal increases in TG levels, with a pooled OR of 1.17 (95% CI: 1.04, 1.33). There was no significant heterogeneity between studies (I2 = 5%, p = 0.37) (Fig. 4-a). The result was meaningless when the study by Fukui N, 2012 was omitted (Supplementary Fig. 3-a). No significant publication bias was found after Egger’s(p = 0.769) and Begg’s tests (p = 1.000).
Periodontitis was a risk factor for abnormal decreases in HDL levels, with a pooled OR of 1.42 (95% CI: 1.24, 1.62, p < 0.01), and there was no significant heterogeneity among the studies (I2 = 0%, p = 0.68) (Fig. 4-b). The result was robust regardless if any one study was omitted (Supplementary Fig. 3-b). No significant publication bias was found after Egger’s(p = 0.282) and Begg’s tests (p = 0.497).
Since the pathological changes in other indicators, including TC, LDL and VLDL, are often not regarded as classic indicators of dyslipidemia, we only analysed the results of hyper TG and low LDL.
Pd And Dyslipidemia
Eight studies evaluated the difference in the PD level between dyslipidemia patients and healthy control groups with the mean (SD). The PD levels were higher in dyslipidemia patients than in controls, with a pooled mean difference of 0.41 mm (95%-CI: 0.23, 0.58, p < 0.01). There was significant heterogeneity between the studies (I2 = 66%, p < 0.01) (Fig. 5-a). Meta-regression showed that the sources of the heterogeneity may include the year of publication (p = 0.038) and region (p = 0.038) (Table S6). The result was robust regardless if any one study was omitted (Supplementary Fig. 4-a). No significant publication bias was found after Egger’s(p = 0.178) and Begg’s tests (p = 0.095).
Cal And Dyslipidemia
Eight studies evaluated the difference in the CAL level between dyslipidemia patients and healthy control groups with the mean (SD). The CAL levels were higher in dyslipidemia patients, with a pooled mean difference of 0.56 mm (95%-CI: 0.35, 0.78, p < 0.01). There was significant heterogeneity between studies (I2 = 62%; p < 0.01) (Fig. 5-b). However, no significant source of heterogeneity was found through the meta-regression (Table S6). The result was robust regardless if any one study was omitted (Supplementary Fig. 4-b). No significant publication bias was found after Egger’s (p = 0.519) and Begg’s tests (p = 0.532).
Bop And Dyslipidemia
Four studies evaluated the difference in the BOP level between dyslipidemia patients and healthy control groups with the mean (SD). No significant difference in BOP levels was found between dyslipidemia patients and healthy controls. There was significant heterogeneity between studies (I2 = 97%; p < 0.01) (Fig. 5-c). Meta-regression showed that the sources of the heterogeneity may be the year of publication (p < 0.001), total sample size (p = 0.004), age (p < 0.001) and sex ratio (p = 0.002) (Table S6). The result was significant when the study by Lutfioglu M, 2017 was omitted (Supplementary Fig. 4-c). No significant publication bias was found after Egger’s(p = 0.848) and Begg’s tests (p = 0.497).
Pi And Dyslipidemia
Six studies evaluated the difference in the PI level between dyslipidemia patients and healthy control groups with the mean (SD). PI levels were higher in dyslipidemia patients, with a pooled mean difference of 0.27 (95%-CI: 0.07, 0.47, p < 0.01). There was significant heterogeneity between studies (I2 = 82%; p < 0.01) (Fig. 5-d). Meta-regression showed that the sources of the heterogeneity may be sex (p = 0.013) (Table S6). The result was meaningless when the study of Dogan B, 2015 or Shivakumar T, 2013 was omitted (Supplementary Fig. 4-d). No significant publication bias was found after Egger’s(p = 0.379) and Begg’s tests (p = 0.453).
Gi And Dyslipidemia
Four studies evaluated the difference in the GI level between dyslipidemia patients and healthy control groups with a mean (SD). No significant difference in GI level was found between dyslipidemia patients and healthy control groups (Fig. 5-e). Meta-regression showed that the sources of the heterogeneity may be Study design(p < 0.001), Quality(p < 0.001) and Gender ratio(p < 0.001) (Table S6). The result was robust regardless if any one study was omitted (Supplementary Fig. 4-e). No significant publication bias was found after Egger’s(p = 0.193) and Begg’s tests (p = 0.050).
Effect Of Periodontal Treatment On Blood Lipids
Three studies evaluated the difference in the association between non-surgical periodontal treatment groups and the control groups with the mean (SD). No difference in other blood lipid level except LDL at baseline (Fig. 6-a-d1). No significant publication bias was found after Egger’s and Begg’s tests (p > 0.05).
Nonsurgical Periodontal Treatment And Tc
Compared with the control group, the level of TC in the serum of patients who received a non-surgical periodontal treatment was decreased significantly after three months, and the pooled mean difference for TC in the treatment and control groups was − 8.32 mg/dL (95% CI: -16.59, -0.05, p = 0.05). There was no significant heterogeneity between the studies (I2 = 0, p = 0.75) (Fig. 6-a2). The result was meaningless regardless if any one study was omitted (Supplementary Fig. 5-a2).
Several studies that reported positive results were excluded from the meta-analysis because they did not have the standardized clinical data we needed. DUAN Jinyu et al. reported that three months after a nonsurgical periodontal treatment, the cholesterol levels were significantly reduced. With 5.72 mmol/l as the diagnostic criterion, four of eight hypercholesterolemia patients returned to normal serum cholesterol levels (17). The research by A. Surdumacove produced similar results; compared with the control group that received only oral hygiene guidance, the test group that received a non-surgical periodontal treatment exhibited a significant decrease in TC levels after one month (18). Zuza EP et al. reported an interesting result: after non-surgical periodontal treatments, TC levels in obese patients were significantly reduced three months later, but the same results were not observed in nonobese patients (19).
Nonsurgical Periodontal Treatment And Tg
Compared with the control group, the level of TG in the serum of patients who received the non-surgical periodontal treatment was decreased significantly after three months, with a pooled mean difference of -36.13 mmol/L (95% CI: -53.63, -18.62, p < 0.01). There was no significant heterogeneity between the studies (I2 = 0, p = 0.77) (Fig. 6-b2). The result was robust regardless if any one study was omitted (Supplementary Fig. 5-b2).
Considering the results of other studies, with 1.70 mmol/l as the diagnostic criterion, DUAN Jinyu reported that the serum cholesterol levels in five of fifteen hypertriglyceridaemia patients returned to normal after the non-surgical periodontal treatment. The observation period was three months (18). This article was not included in the meta-analysis because there were no specific parameters. Zuza EP also reported similar results (19).
Nonsurgical Periodontal Treatment And Hdl
Compared with the control group, the level of HDL in the serum of patients who received the non-surgical periodontal treatment was increased significantly after three months, with a pooled mean difference of 3.98 mmol/L (95% CI: 1.71, 6.25, p < 0.01). There was no significant heterogeneity between the studies (I2 = 0, p = 0.81) (Fig. 6-c2). The result was meaningless when the study by Fu YW, 2016 was omitted (Supplementary Fig. 5-c2).
Nonsurgical Periodontal Treatment And Ldl
Finally, we performed a meta-analysis of the LDL levels in serum. Analysis of these studies showed that there was no statistically significant difference in the LDL levels between the treatment and control groups after three months of treatment (Fig. 6-d2). The result was significant when the study by Fu YW, 2016 was omitted (Supplementary Fig. 5-a).
Effect Of Lipid Treatment On Periodontitis
Five studies evaluated the difference in the association between the lipid treatment and periodontitis with the mean (SD). No significant publication bias was found after Egger’s and Begg’s tests (p > 0.05).
Compared with that of the control group, the level of GI in the dyslipidemia patients who received the lipid treatment decreased significantly, with a pooled mean difference of -0.15 (95%-CI: -0.25, -0.06, p < 0.01). There was no significant heterogeneity between the studies (I2 = 0, p = 0.92) (Fig. 7-e).
We found no statistically significant difference in PD, CAL, BOP, or PI between the treatment and control groups (Fig. 7). Through meta-regression, it was determined that the sources of the heterogeneity may be the total sample size for PD (p = 0.017), study design for CAL (p = 0.007), and age for PI (p = 0.028) (Table S7).
The following results are reported in related studies that are not included in the forest map. Özlem FENTOĞLU reported that two months after the periodontal treatment and lipid treatment, PI, GI, BOP, and PD in the statin treatment group were significantly reduced, while similar results were observed in the diet control group. (20)