Study design and ethical consideration
This cross-sectional study randomly selected participants from public advertisement. All of the participants voluntarily provided a written informed consent. The Institutional Review Board for Human Subjects of Seoul National University Dental Hospital reviewed and granted the ethical consideration for this study (CRI17009).
Sample size estimation
The results of pilot test using ten participants (each five cases and controls) showed that S100A8 in saliva (mean± standard deviation [SD], ng/ml) was 8.0 ± 10.2 for periodontitis patients versus 6.1 ± 8.3 for healthy participants. Under the condition of type I error of 0.05, type II error of 0.8 and ratio of 2 between periodontitis patients and healthy participants, the sample size of this study was estimated as total of 147 (98 periodontitis patients and 49 healthy participants).
The following inclusion criteria were set for this study, (1) agreed to have periodontal examination including clinical attachment loss (CAL) and panoramic radiograph according to the new international periodontal classification guideline by American Academy of Periodontitis and European Federation of Periodontology , (2) aged over 20 years, (3) having no medication during previous three months (4) having idea to donate adequate sample of blood, GCF and saliva for analysis (5) having no missing data used in the final analysis.
Assessment of periodontitis
CAL and radiographic bone loss of each natural tooth were examined by trained dentists using panoramic radiograph (Pax-Primo, Vatech Global, Seoul, Korea). CAL was measured in all of the natural teeth except 3rd molar by using a UNC-15 probe. Periodontal status was categorised according to the guidelines of 2017 AAP-EFP workshop in Periodontology . The participants diagnosed as Stage II-IV periodontitis were considered as periodontitis (‘periodontitis’) and the other participants with no or Stage I periodontitis were non-periodontitis (‘healthy’). Stage II-IV periodontitis is CAL ≥3mm or extraction due to periodontitis or radiological bone loss>15% of coronal third .
Assessment of clinical periodontal parameters
Plaque index (PI), pocket depth (PD) and bleeding on probing (BOP) were considered as periodontal clinical parameters. PI was evaluated by Turesky modification of the Quigley-Hein Index  . PD was evaluated at six sites per tooth (mesio-, mid- and disto- buccal and lingual) using a UNC-15 probe and dichotomized according to PD≥ 4mm. BOP was evaluated using the guideline in a previous study . Finally, PI, PD and BOP were presented as PI, percentage of site with PD ≥ 4mm and percentage of site with BOP positive.
Each participant has information about standard sampling protocol for saliva collection. Patients have information that they do not have to brush tooth, drink or eat one hour before sampling. Unstimulated whole saliva was collected using passive drooling method for 10 minutes in a 50ml conical tube in order to maintain consistency of samples. Collected saliva in tubes was centrifuged (2600 x g for 15 minutes at 4ºC) and supernatants were iquoted into 1 ml in sterilised 1.5 ml Eppendorf tube. The tubes with saliva samples were stored at -80 ºC for further analysis.
Radiographic evaluation and periodontal probing using UNC-15 probe were applied to decide the deepest pocket among all teeth pockets. GCF samples were obtained from the deepest pocket. During GCF sampling, we tried to avoid blood and saliva contamination. After isolating the tooth with cotton rolls, three absorbent paper points (#25, Meta Biomed Inc., Chungbuk, Korea) were gently inserted in the deepest pocket for 30 seconds. Paper points were immediately placed in a cryovial containing 1 ml of phosphate buffer saline (PBS) in pH 7.4, which were centrifuged (2600 x g for 15 minutes at 4ºC) and supernatants were aliquoted into 1 ml in sterilised 1.5 ml Eppendorf tube. The tubes with GCF samples were stored at stored at -80 °C for further analysis.
Blood of 4 ml was drawn by venepuncture by a trained medical technologist. The blood samples were centrifuged (2600 x g for 15 minutes at 4ºC) and supernatants (plasma) were aliquoted into 1 ml in sterilised 1.5 ml Eppendorf tube. The tubes were then stored at -80 °C for further analysis.
Quantification of salivary S100A8 and S100A9
S100A8 and S100A9 protein concentrations were determined from saliva, blood and GCF using enzyme-linked immunosorbent assay (ELISA) kit (R&D systems, Minneapolis, MN, USA) following manufacturer’s instruction. Standard curve was drawn using standard S100A8 and S100A9 supplied by the manufacturer. GCF and saliva samples were diluted on concentration dependent using reagent diluent provided by manufacture (1, 1/2, 1/4, 1/8, 1/16, 1/32) and diluted sample concertation for S100A8 and S100A9 were calculated from standard curve of S100A8 and S100A9. Similarly, blood samples were diluted on concentration dependent manner with reagent diluent (1/10, 1/20, 1/40, 1/80, 1/160, 1/320) and concentrations of S100A8 and S100A9 were estimated using standard curve. We decided the standard dilution rate that falls on the range of 500 pg/ml to 1000 pg/ml on pilot study.
Assessment of confounding variables
Sociodemographic factors (sex and education), behavioural factors (smoking and drinking) and systemic health information (diabetes) hypercholesterolemia, hypertension and obesity) were considered as confounders. Data for all confounders were collected using face to face interview, laboratory blood analysis and physical examination. Dichotomised variables were as follows, sex: male and female, education: until middle school and above high school, smoking: smoker and non-smoker, drinking: alcohol drinker and non-drinker. Four systemic health components were as follows: 1) diabetes: high plasma glucose level (>126 g/dl) or having anti-diabetic medication, 2) hypercholesterolemia: high plasma cholesterol level (>240 mg/dl) or having anti-hypercholesterolemia medication, 3) hypertension: systolic >130 mmHg or diastolic >85 mmHg or having anti-hypertensive medication and 4) obesity: body mass index (BMI) calculated as kg of body weight divided by square meter of height ≥25. Physicians measure the blood pressure and diagnosed the hypertension. The blood pressure was measure in the sitting position using mercury manometer. For biochemical variable, 12 hour fasting blood samples were drawn at recruitment.
The distribution of characteristic variables by periodontitis (no versus yes) were addressed using mean values with standard deviations (SD) for continuous variables, and frequencies and proportion for categorical variables. Chi-square test was applied for categorical variables. Kolmogorov-Smirnov (K-S) test was applied to evaluate the normal distribution for continuous variables. When variables were in normal distribution, parametric tests were applied, otherwise non-parametric tests were applied. T-test were performed to evaluate difference in continuous variables with normal distribution. Mann-Whitney (M-W) test were applied for continuous variables without normal distribution. The relationships between values in the blood, GCF and saliva were analysed with Spearman’s correlation test. Since number of participants were big (n=149), Analysis of covariance (ANCOVA) was applied to estimate adjusted mean with standard error (SE) of S100A8 and S100A9 levels after controlling for age, sex, education, smoking, drinking, diabetes, hypercholesterolemia, hypertension and obesity. The receiver operating characteristic (ROC) curve was applied for estimating c-statistics (area under the curve: AUC) as screening ability of S100A8 and S100A9 for periodontitis. Statistical significance was set at p-value <0.05. Data were analysed using Statistical Package for Social Sciences version 25 (SPSS inc, Chicago, Il, USA).