Study design

This cross-sectional study was conducted in accordance with the Declaration of Helsinki on Human Studies. All experimental protocols involving patients and healthy individual samples were approved by the Ethics Committee of Kyushu Dental University (No. 18–32). Data collection was conducted according to the Strengthening the Reporting of Observational Studies in Epidemiology guidelines.

We enrolled a cohort of 62 patients with a total of 137 implants who received maintenance care at Kyushu Dental University Hospital between July 2022 and November 2023. The inclusion criteria were: (1) age ≥ 20 years, (2) absence of pregnancy or lactation, (3) use of functioning implants for at least 12 months, (4) no history of poorly controlled systemic diseases, (5) no history of nonsurgical or surgical treatment, such as scaling at the site, to be examined within 3 months of examination, and (6) no history of medical treatment during the last 3 months before examination and sampling.

The exclusion criteria were: (1) presence of implants without previous radiographs (base data), (2) presence of implants placed in a position or with superstructures that made probing difficult, (3) presence of implants with an average marginal bone loss of ≥ 0.2 mm and BOP (-), and (4) presence of implants for which the amount of bone resorption could not be measured due to unclear radiographic images.

Ultimately, 29 patients (15 males and 14 females; mean age: 74.6 years) with a total of 76 implants were included for final analyses. All participants provided informed consent.

Clinical evaluations

The participants’ implants were assessed by a trained dentist (YS) using the following measurements: Peri-implant conditions were evaluated using a plastic probe (Colorvue, Hu-Friedy, Chicago, IL, USA) under low pressure (0.25 N) for PPD, presence of BOP, modified plaque index (mPI), and modified gingival index (mGI) [22]. Radiographic assessment involved random assignment of radiographs to each evaluator (TN, TM, and TM), with evaluators blinded to any patient-identifiable information. Radiographic examination was conducted using distance measurement software (VHX-5000, Keyence, Tokyo, Japan) in an electron microscope to measure the distance between the proximal bone junction of the implant and the most apical side of the implant, with the implant shoulder serving as the reference point. Subsequently, we calculated the average of the measurements taken by the three evaluators. Marginal bone loss (MBL) [23], was calculated by adjusting values for the magnification ratio of the length of the implant body. The average annual bone loss (ABL) around the implants was then calculated and compared with the MBL at the baseline (Fig. 1a and b).

Fig. 1

(a) Bone resorption measurement using radiographic images. The distance between the proximal bone junction and the most apical side of the implant was measured using the implant shoulder as a reference point. The resulting values were corrected by the magnification ratio to the long diameter of the implant body, and marginal bone loss (MBL) was calculated. (b) Formula for calculating average annual bone loss (ABL). The ABL around the implants was calculated in comparison to the MBL at the baseline. (c) The PISF was obtained using PerioPaper from the site where the PPD was deepest

Patient groups

Following the 2017 World Workshop [7] guidelines, the 76 eligible implants were categorized into three groups: healthy (n = 29), mucositis (n = 22), and peri-implantitis (n = 25), defined by the characteristics presented below. Pathological bone resorption was defined as an ABL of ≥ 0.2 mm at the peri-implant area, according to the Toronto Conference [24].

1)

Healthy group (peri-implant health): Implants with BOP (-), ABL < 0.2 mm, and no other signs of inflammatory lesions on the oral mucosa.

2)

Peri-implant mucositis group (peri-implant mucositis): Implants with BOP (+) and ABL < 0.2 mm.

3)

Peri-implantitis group: Implants with BOP (+) and ABL > 0.2 mm.

PISF sampling

Plastic curettes (Implacare; Hu-Friedy, Chicago, IL, USA) were used to remove plaque above the peri-implant margin. Sampling sites were isolated using cotton rolls and dried using a gentle stream of air. PerioPaper® (OraFlow Inc.; Plainview, NY, USA) was gently inserted < 1–2 mm into the deepest sulcus until a slight resistance was felt, and then held in place for 1 min (Fig. 1c). Samples were collected five times from the same site using the same method, with a 1-minute interval between each collection. Any PerioPaper contaminated with blood or saliva was discarded and replaced after 10 min. To minimize evaporation, volume quantification was performed immediately after sampling using a Periotron 8000 device (OraFlow Inc.). The Periotron 8000 was calibrated prior to the study and recalibrated periodically, following the manufacturer’s instructions. Periotron values are expressed as the volume of PISF (µL) with reference to the corresponding calibrated logarithmic curve [25]. PerioPaper was stored in a 50 µL mixture of phosphate-buffered saline (PBS) and protease inhibitors (cOmplete™, Sigma-Aldrich, St. Louis, MO, USA) in plastic sealable Eppendorf tubes and frozen at -80 °C until analysis.

Enzyme-linked immunosorbent assay

The solution collected using PerioPaper was vortexed for 10 min to elute and then centrifuged at 300 rpm for 10 min at 4 °C. Subsequently, centrifugation was performed at 12,000 rpm for 2 min. The resulting supernatant was collected, and five supernatants were combined to yield a total volume of 250 µL. ET-1 levels were measured using the Quantikine® Enzyme-linked immunosorbent assay (ELISA) Endothelin-1 Immunoassay kit (R&D Systems, MN, USA), while IL-1β levels were measured using the Quantikine® ELISA Huma IL-1β/IL-1F2 kit (R&D Systems). ELISA procedures were performed according to the manufacturer’s instructions. Sites with cytokine concentrations below the detection limit of the assay were recorded as 0. These biomarker concentrations were adjusted for the amount of PISF and expressed as ET-1 (ρg/site) and IL-1β (µg/site) [26].

Statistical analysis

G*Power 3.1.9.6 software was utilized for sample size calculations, with an effect size of 0.8, a statistical power of 80%, and a significance level of 95% (α < 0.05), two-tailed. Based on these parameters, a minimum of 21 implants were required for each group to detect a difference between the groups, which served as the sample size requirement of the study. Statistical analyses were performed using Bell Curve for Excel (Social Survey Research Information Co., Ltd., Tokyo, Japan). Data normality was evaluated using the Shapiro–Wilk test. The Kruskal–Wallis test was employed to determine statistically significant differences between groups for each independent variable, with subsequent Steel–Dwass adjustment. The diagnostic accuracy of the biomarker candidates for distinguishing peri-implantitis or peri-implant mucositis from healthy implants was evaluated using receiver operating characteristic (ROC) curve analysis and the area under the ROC curve (AUC). Each biomarker, along with its adjusted logistic regression model (adjusted for sex and age of the dental implant), was adjusted. The Youden index was utilized to determine the optimal cut-offs from the ROC curves for each biomarker (unadjusted and adjusted models). Diagnostic sensitivity and specificity were calculated for each biomarker using a cut-off value to assess classification quality. Statistical significance was set at p < 0.05.