Polymorphisms in Genes of Lipid Metabolism Are Associated with Type 2 Diabetes Mellitus and Periodontitis, as Comorbidities, and with the Subjects' Periodontal, Glycemic, and Lipid Profiles

Background Type 2 diabetes mellitus (T2DM) and periodontitis (P) commonly occur as comorbidities, but the commonalities in the genetic makeup of affected individuals is largely unknown. Since dyslipidemia is a frequent condition in these individuals, we investigate the association of genomic variations in genes involved in lipid metabolism with periodontal, glycemic, lipid profiles, and the association with periodontitis and T2DM (as comorbidities). Methods Based on clinical periodontal examination and biochemical evaluation, 893 subjects were divided into T2DM+P (T2DM subjects also affected by periodontitis, n = 205), periodontitis (n = 345), and healthy (n = 343). Fourteen single-nucleotide polymorphisms (SNPs) were investigated: LDLR gene (rs5925 and rs688), APOB (rs676210, rs1042031, and rs693), ABCC8 (rs6544718 and 6544713), LPL (rs28524, rs3735964, and rs1370225), HNF1A (rs2650000), APOE (rs429358 and rs7412), and HNF4A (rs1800961). Multiple linear and logistic regressions (adjusted for covariates) were made for all populations and stratified by sex and smoking habits. Results Individuals carrying APOB-rs1042031-CT (mainly women and never smokers) had a lower risk of developing periodontitis and T2DM (T2DM+P); altogether, this genotype was related with healthier glycemic, lipid, and periodontal parameters. Significant disease-phenotype associations with gene-sex interaction were also found for carriers of APOB-rs1676210-AG, HNF4A-rs1800961-CT, ABCC8-rs6544718-CT, LPL-rs13702-CC, and LPL-rs285-CT. Conclusions Polymorphisms in lipid metabolism genes are associated with susceptibility to T2DM-periodontitis comorbidities, demonstrating gene-sex interaction. The APOB-rs1042031 was the most relevant gene marker related to glucose and lipid metabolism profiles, as well as with obesity and periodontitis.


Introduction
Periodontitis (P) is characterized by dysbiosis of the commensal microbiota associated with inflammation [1], which is responsible for the destruction of the hard and soft periodontal tissues that surround the tooth (connective gingival and alveolar bone tissue, cementum, periodontal ligament) causing dental mobility and culminating in tooth loss [2] It is considered the main cause of tooth loss in adults, affecting approximately 10-15% of the world population [3].
It derives from a complex and dynamic interplay among the immune system, microbiota, and lifestyle habits (e.g., smoking, and stress) ultimately resulting in homeostatic adaptation and maintenance of clinical health or disease manifestation. This interplay involves a highly complex network of host-derived molecules, which is largely determined by individual genomic variations. Individual genetic variations partly explain why manifestation of P often lacks correspondence with quality and amount of oral microbiota and also differs markedly between individuals with similar environmental backgrounds and comparable lifestyle habits [4].
Diabetes mellitus (DM) is a metabolic disorder that involves persistent hyperglycemia, resulting from defects in insulin production, insulin action, or both [5]. Type 2 diabetes mellitus (T2DM) occurs when there is a progressive loss of insulin secretion combined with insulin resistance [6,7]. Overwhelming evidence indicates that T2DM is associated with increased prevalence and severity of P [8]. In fact, P is currently acknowledged as the sixth "classical" complication of T2DM [9].
Dyslipidemia is a deregulation of lipid metabolism, characterized by qualitative and/or quantitative changes in plasma lipoproteins and lipids [10]. Dyslipidemia is highly prevalent in T2DM [11], and T2DM is considered to cause dyslipidemia [12] and T2DM and dyslipidemia are closely related to cardiovascular complications, an important cause of morbidity and mortality [13]. A meta-analysis [14] supports the association between P and dyslipidemia, as the occurrence of P is significantly associated with reduced HDL levels and elevated serum concentrations of LDL and triglycerides. Thus, it is conceivable that dyslipidemia has a role mediating the interaction between T2DM and P, increasing the risks and severity of both conditions [15].
Periodontitis, T2DM, and dyslipidemia are complex conditions, which have in common their association with a systemic hyperinflammatory state [16]. In multifactorial diseases, genetic, epigenetic, and environmental factors play an essential role in their pathogenesis [17][18][19]. More studies are necessary to better understand the potential connection among P, T2DM, and dyslipidemia. Considering that the literature demonstrates a variety of genes relevant for dyslipidemia [20][21][22][23][24][25][26][27][28], we hypothesize that polymorphisms in genes of lipid metabolism could impact the occurrence of periodontitis alone or together with T2DM. Therefore, this study was aimed at investigating whether polymorphisms in lipid metabolism genes are associated with T2DM and periodontitis concomitantly present as comorbidities. Also, we investigate if specific genetic variants in these lipid metabolism genes are associated with glycemic and lipid profiles, as well with the periodontal status.

Materials and Methods
2.1. Study Subjects, Sample Collection, and DNA Isolation. A sample of 893 individuals with no familiar relations were included in the study from 1158 patients screened from the population who attended the Clinics of Periodontology, School of Dentistry at Araraquara, São Paulo State University (UNESP), Brazil, from 2010 to 2018. Subjects were excluded if they were younger than 30 years old, had chronic or systemic diseases (i.e., HIV infection or immunosuppressive chemotherapy), had received systemic antibiotics in the previous 3 months, were subjected to periodontal treatment in the previous 6 months, had less than 10 natural teeth (excluding third molars), and/or were pregnant or lactating.
The sample size was calculated with the G * Power Calculator, version 3.1.9 [29], as detailed in the Supplementary material. This study was approved by the Ethics in Human Research Committee of the School of Dentistry at Araraquara (UNESP; Protocol number 50/06) and followed the guidelines of the Helsinki Declaration of 1975 (revised in 1983).
A complete periodontal examination of all subjects assessed the following clinical signs and parameters at four interproximal sites around each tooth: probing pocket depth (PPDi) and clinical attachment level (CALi), measured to the nearest millimeter by a UNC-15 periodontal probe (Millennium®) and the presence of visible plaque, marginal bleeding and bleeding on probing (BOP) [32,33]. The last three parameters were registered as a percentage of the total number of interproximal sites assessed. All fully erupted teeth, except third molars and teeth with extensive destruction of the clinical crown, were examined. The criteria for the definition of periodontitis were according to the CDC/AAP (Centers for Disease Control and Prevention/American Academy of Periodontology) [4].
A specific questionnaire confirmed smoking status. History of smoking was defined according to Miyake et al. as follows: (i) never smoked or (ii) ever smoked (including patients those who are current smokers and former smokers) [34].
Genomic DNA was extracted from buccal mucosa epithelial cells from each patient, using standard protocols involving proteinase K digestion, 8 M ammonium acetate, and subsequent ethanol precipitation [35]. The purity of genomic DNA was assessed on a microvolume UV spectrophotometer (NanoDrop® 2000, Thermo Scientific) and quantified on a fluorometer (Qubit® 2.0 fluorometer, Invitrogen). Samples were used whether they presented A 260/280 ratio between 1.7 and 2.0, and the concentration was adjusted to 50 ng/μL in TE buffer (Tris-HCl 10 mM, EDTA 0.1 M, SDS 0.5%-pH 8.0).
The genotyping analysis was performed with an online software (Applied Biosystems™ Analysis Software, Genotyping Analysis Module, version 3.2.).

Statistical Analysis.
The general characteristics of each group were submitted to descriptive and analytical statistics using IBM SPSS Statistics 20.0 and GraphPad Prism 6.0. First, all data were evaluated by normality tests (D'Agostino-Pearson); then, they were applied with parametric (multiple comparisons with the one-way ANOVA followed by Tukey's test, expressed as the mean [±standard deviation]) or nonparametric (multiple comparisons by Kruskal-Wallis test followed by Dunn's test, expressed as the median minimum-maximum]) tests. Continuous demographic variables were compared among the groups using Student's t -test (for age). The chi-squared test (X 2 ) was used for categorical variables (sex, ethnicity, and smoking).
The Hardy-Weinberg equilibrium (HWE) formula was applied to the genotype frequencies, separately for each group, using chi-square goodness-of-fit for any deviations from the expected genotype equilibrium. A chi-squared test (X 2 ) was also used to assess differences in genotypic and allelic distributions of each SNP between the groups of patients. All these analyses and the minor allele frequency (MAF) of each SNP were calculated using the PLINK software [36].
Multiple logistic regression models assuming additive allelic effects, adjusting for age, sex and smoking habits (as independent variables) were used for testing associations between SNPs and the occurrence of only P (when comparing healthy and periodontitis groups), the occurrence of only T2DM (when comparing periodontitis and T2DM+P groups), and the occurrence of T2DM together with P (when comparing healthy and T2DM+P groups). These gene-phenotype associations were estimated by the odds ratio (OR) and confidence interval (CI) of 95%. The distributions of genotypes and alleles were compared in the two groups by chi-squared tests, and differences were considered significant when p < 0:05. Furthermore, the same analysis was stratified by sex to identify the sex-specific effects of the SNPs on periodontitis by the additive model using multiple logistic regression adjusted by age and smoking. Similarly, to identify the effects of smoking habits with those SNPs on periodontitis, the population was stratified by smoking habits followed by the multiple logistic regressions adjusting by age and sex. Differences were considered significant when p < 0:05, but whether the Bonferroni correction is considered, the experiment-wise significance is the p value < 0.003, as determined by the total number of genotyped markers in the study (0.05/14 SNPs).
Spearman's correlations among glycemic, lipid, and periodontal parameters were made for each group (Supplementary Figure 1). A multiple linear regression was used to assess the relationship of each SNP independently with periodontal parameters (number of remaining teeth, marginal bleeding, BOPi, PPDi, and CALi) and with glycemic, lipid, and obesity parameters (fasting glucose, insulin level, HbA1c, triglycerides, total cholesterol, HDL, LDL, BMI, and waist-to-hip ratio). All these analyses and multiple logistic regressions were performed using the STATA software version 12.0 for Windows (Statistics/Data Analysis, Stata Corporation, College Station, TX, USA).

Results and Discussion
3.1. Results. Regarding the demographic characteristics, there were statistical differences among the groups in the mean of age, sex, and smoking habits. Healthy individuals were younger (mean 43 y.o.) ( Table 2). Most of the investigated population was composed by female and never- Total cholesterol and triglycerides levels were higher in the T2DM+P group, as well as the body mass index (BMI) and waist-to-hip ratio (Table 2). Collectively, these results support an association or clustering of obesity/dyslipidemia/T2DM/periodontitis in this studied population.
In the Supplementary Table 1, besides the allelic and genotypic distributions among the groups, it is possible to note that the majority of SNPs presented a minor allele frequency (MAF) greater than 0.05. The genotype distributions of each SNP did not deviate from the Hardy-Weinberg equilibrium in the healthy group.
Additional result analyses in the different groups are presented in the Supplementary material. Supplementary  Figure 1 shows Spearman's correlations of biochemical and periodontal data in each studied group. In Supplementary  Table 2, we present data of median (minimum-maximum) of glycemic and lipid profiles and physical exams of individuals belonging to the T2DM+P subgroups. In this Supplementary Table 2, when we subdivided T2DM+P by the severity of periodontitis, and we compared wellcontrolled T2DM+P with poorly controlled T2DM+P, we observed that the patients with severe periodontitis presented, in general, a higher glycemic and lipid biochemical levels. For example, poorly controlled with severe periodontitis T2DM+P patients showed the highest levels of fasting glucose and HbA1c.
When we considered T2DM_P moderate versus severe periodontitis patients compared by their lipid parameters, we observed in dyslipidemic T2DM patients with severe P statistically significant higher levels of total cholesterol, LDL-cholesterol, and triglycerides, suggesting that the severity of P could be related with the worst lipid profile of the patients.
Multiple linear regression analyses indicate that the APOB-rs1042031-CT was related to the number of remaining teeth (β = −0:08; p = 0:017), bleeding on probing (β = −0:05; p = 0:03), PPDi ≥ 5 mm (β = −0:08; p = 0:008), and CALi ≥ 6 mm (β = −0:07; p = 0:01) ( Table 4). A negative coefficient suggests that as the independent variable increases, the dependent variable tends to decrease, thus supporting a possible protective role for this SNP in periodontitis. In contrast, the rs676210-AG SNP in the same APOB gene was positively associated with bleeding on probing (β = 0:05; p = 0:02) and CALi ≥ 6 mm (β = 0:07; p = 0:01 ). The SNP rs5925 (LDLR gene) was significantly related to marginal bleeding (β = −0:17; p = 0:02), which may suggest an influence on immune response to oral biofilm bacteria. Multiple linear regression analyses assessing the association of each SNP with biochemical and obesity parameters ( Table 5) also detected the APOB-rs1042031-CT SNP as relevant for glycemic, lipid, and obesity parameters, particularly the waist-to-hip ratio (β = −0:25; p ≤ 0:01). The rs676210-AA SNP in the same APOB gene was positively related to fasting glucose (β = 0:21; p = 0:03), supporting a role for this genetic variation in T2DM. The rs285 SNP (LPL gene) was also related to BMI and to waist-to-hip ratio ( Table 5). Results of Spearman's correlations between biochemical profiles and periodontal parameters for each group can be found in the Supplementary Figure 1 3.2. Discussion. This case-control study demonstrates the association of polymorphisms in genes relevant for lipid metabolism with the occurrence of periodontitis (P) or the occurrence of T2DM and P concomitantly (T2DM+P). Specifically, the rs1042031 SNP in the APOB gene and the rs6544718 SNP in the ABCC8 gene were protective, as their presence is associated with a significantly lower risk of T2DM+P. In contrast, the rs676210 (APOB), rs13702 (LPL), and rs1800961 (HNF4A) SNPs were associated with increased risk of T2DM+P. There was no significant association between SNPs in LDLR and APOE genes with either P or T2DM+P in Brazilian subjects.
Polymorphisms in the APOB gene have been associated with variability in serum cholesterol levels and coronary atherosclerosis in some populations [37,38]. The rs1042031 SNP involves a silent mutation from cytosine to thymine       [22,39,40], osteonecrosis [41], and ischemic stroke [23]. Here, we identified the association of APOB-rs1042031-CT with the lower chance of presenting P and T2DM concomitantly (T2DM+P). After the sample stratification by sex or smoking habits, we confirm these findings in females and in never-smoking individuals. Subsequent stratification by gender and smoking status (together) confirmed that APOB-rs1042031-CT females who never smoked had a lower chance to develop T2DM+P (p = 0:004). It was reported that the risk of having plaques on carotid arteries was lower in T2DM subjects carrying mutant specific alleles (E-alleles) [42]. The rs676210 SNP is a missense mutation involving G to A substitution that results in the conversion of proline to leucine, also named as Pro2739Leu. The rare allele A at the rs676210 SNP was significantly associated with the increased triglyceridelowering effects of fenofibrate, a drug commonly used to           [43], with reduced levels of LDL and triglycerides in Europeans [44]; and GWAS also associated this mutation with increased circulating oxidized LDL (oxLDL) [45]. We identified significant association of the APOB-rs676210-AG with presence of T2DM+P, whereas the same SNP in homozygosis (rs676210-AA) was associated with increased fasting glucose (OR = 0:21; 95%CI = 0:01 − 0:41; p = 0:03, Table 5) in our population.
Another relevant result found here was that the poorly controlled with severe periodontitis T2DM+P patients showed the highest levels of fasting glucose and HbA1c. It is worth to take into mind that 40% of the given plasma HbA1c pool reflects glycemia levels of the previous 31-90 days and 10% during the previous 91-120 days [46,47] and that hyperglycemia inhibits new bone formation and exacerbates alveolar bone resorption due to periodontal disease [48].
Genetic variants in the ABCC8 gene (OMIM: 600509) have also been widely explored in candidate genes casecontrol studies due to its function of encoding the sulfonylurea receptor, which plays an important role in insulin secretion [49]. Pharmacogenetics studies usually investigate this gene because of its relevance for the efficacy of hypoglycemic drugs, since the failure in oral drug therapy leads to poorly regulated glycemia and dyslipidemia [24,50]. Thus, due to the relevance of ABCC8 gene for glycemia in patients with T2DM, it is plausible that this gene may also affect the occurrence and severity of periodontitis in these patients, via poor metabolic control. From the two SNPs investigated in the ABCC8 gene, rs6544718 showed a gender-biased positive association with a lower chance to develop T2DM+P, as men carrying rs6544718-CT were 63% less likely to develop T2DM+P. Another gender-biased genetic influence was observed for the LPL gene (OMIM 609708), with men carrying the rs13702-CC SNP presenting increased occurrence of T2DM+P. Interestingly, in a study enrolling patients affected by only T2DM, Hatefi et al. [25] reported the relationship between this CC genotype with a greater susceptibility to develop T2DM, in agreement with our results. The rs13702-CC may favor the entry of fatty acids into the liver, adipocytes, and skeletal muscle, which leads to the development of insulin resistance and, later, T2DM [25]. Another SNP in the same LPL gene, the rs285-CT, was associated with a lower chance of developing periodontitis. Moreover, the LPL-rs285-CC SNP was also related to the waist-to-hip ratio, as well as to BMI (Table 5).
Polymorphisms in the hepatocyte nuclear factor-4α (HNF4A) (OMIM: 60028) and HNF1 homeobox A (HNF1A) (OMIM 14210) genes were included in the present study because both genes encode important transcription factors involved in the regulation of lipid and glycemic metabolism [28]. HNF4A has been extensively associated with maturity-onset diabetes of the young (MODY) characterized by deficient pancreatic beta cell insulin secretion [51,52], which was recently accurately revised by Banday et al. [53]. HNF1A is a transcription factor that interacts with HNF4A and leads to subtypes of MODY and, consequently, increases T2DM susceptibility. This gene resides in the T2DM-linked region on the chromosome 20 which supports a role as a positional candidate gene for the disease [51]. HNF4A-regulated proteins are involved in the pathogenesis of T2DM [54]. Patients with T2DM have been shown to have a characteristic deregulation of lipid metabolism, with lower levels of very low-density lipoprotein (VLDL) [26]. We identified that HNF4A-rs1800961-CT increases the risk of T2DM+P almost twofold. These patients are heterozygotes for the rare nonsynonymous mutation T130I (rs1800961). Other SNPs in the HNF4A gene were associated with elevated serum lipid levels and presence of metabolic syndrome, characterized by elevated glucose parameters in dyslipidemic Finnish and Mexican families [27].
There is no consensus regarding the sex-specific susceptibility to complex diseases in humans. Some studies had reported that men had the worst periodontal conditions [55,56], and women with low carbohydrate intake were protected against development of T2DM [49]. Clinical [57], in vitro [58], and in vivo [59] studies have indicated the influence of steroid hormones on inflammation, immune response, and bone metabolism, demonstrating an association of periodontitis with sex hormone levels. More studies specifically designed to assess the sex or smoking habits are necessary to investigate the role of these SNPs in the context of the complex diseases here investigated.
The present case-control study was not age-sex matched, and the absence of a group with subjects affected by only T2DM were limitations of this study. Because of this fact, we are not able to associate a SNP with T2DM as an isolated disease. Since the population studied here was a convenience sample of patients seeking care/referred to the periodontics clinic, it was expected that these subjects were affected by periodontitis. Nevertheless, patients diagnosed by T2DM frequently have periodontitis, which may reflect the difficulty in achieving and sustaining proper metabolic control allowing for the influence of diabetes-related changes in angiogenesis, immune response, and repair mechanisms on the host-microbial interactions in the periodontal microenvironment. In fact, evidence supports that in wellcontrolled T2DM, the extent and severity of periodontitis, as well as the response to treatment, are similar to that of systemically healthy patients. Thus, it is possible that the duration of diabetes (i.e., time since diagnosis) and quality of metabolic control over time may have varied wildly in the studied subjects. Although we had mentioned the use of Bonferroni's correction, we did not strictly consider the 95% confidence for statistical significance. This was a compromise to reduce the chance of type II error [60][61][62], which would lead us to reject a true biological association hypothesis. Thus, our results should be interpreted with caution, since some of the p values lower than 0.05 may in fact reflect type I errors (false positives).
Although the sample size allowed for adequate statistical power, it is important to expand the findings on these genetic variants in larger Brazilian populations from different regions of the country and in multiple ethnic populations throughout the world, in agreement with Laine et al. [63]

Conclusions
In conclusion, polymorphisms in genes involved in lipid metabolism are associated with susceptibility to the occurrence of T2DM-Periodontitis, with a gene-sex interaction. The APOB-rs1042031 was the most relevant gene marker related to glucose and lipid metabolism profiles, as well as with clinical obesity and periodontitis. Further studies are necessary to confirm these results, investigating the potential functional role of those SNPs at transcriptional and translational levels, and the potential influence of sex in this context.

Data Availability
The genotypes and biochemical data used to support the findings of this study are included within the article and the supplementary information file.