Dental implants are widely recognized as the highly successful and effective treatment modality for replacing missing teeth, and their use has increased dramatically over the past several decades (Giok et al., 2024). However, along with the widespread use of implants, the occurrence of peri-implant diseases has also risen, leading to growing concern. Peri-implantitis is characterized by inflammation of the peri-implant mucosa accompanied by progressive loss of supporting bone, which can ultimately lead to implant failure if left untreated (Joshi et al., 2025). Recent meta-analyses have reported the prevalence of peri-implantitis to be approximately 22% at the patient level and 9-14% at the implant level, posing a substantial clinical burden (Diaz et al., 2022; Galarraga-Vinueza et al., 2025). The development of peri-implantitis involves various risk factors beyond microbiological factors, including smoking, diabetes, history of periodontitis, poor oral hygiene, and advanced age (Cui et al., 2025). The etiology of peri-implantitis is known to be initiated by microbial biofilms, similar to periodontitis around natural teeth. Early studies reported similarities between the microbial communities of peri-implantitis and periodontitis, with the "Red complex" consisting of Porphyromonas gingivalis, Tannerella forsythia, and Treponema denticola, along with the "Orange complex" including Fusobacterium nucleatum and Prevotella intermedia, identified as major pathogens in both diseases (Chun Giok and Menon, 2023). However, with advances in next-generation sequencing (NGS) technology and deepening understanding of the oral microbiome, recent studies have reported that differences may exist between the microbial communities of peri-implantitis and periodontitis (Yu et al., 2024). In particular, implant surfaces differ from natural tooth surfaces in their physicochemical properties, and these differences may influence biofilm formation and microbial community structure. The roughness, surface energy, and chemical composition of titanium implant surfaces are known to affect initial bacterial attachment and biofilm maturation processes (Sinjab et al., 2024). Additionally, peri-implant tissues differ anatomically and immunologically from periodontal tissues around natural teeth, potentially resulting in different host-microbial interaction patterns (Di Spirito et al., 2024). For these reasons, applying the microbiological characteristics of periodontitis to peri-implantitis may have limitations. Most existing microbiome studies of peri-implantitis have been conducted by collecting samples from peri-implant pockets using paper points or curettes (Anitua et al., 2024; Anuntakarun et al., 2025). This approach has the limitation of potentially capturing more planktonic bacteria within the pocket or bacteria from soft tissue surfaces rather than biofilms directly attached to the implant surface. To accurately understand the characteristics of pathogenic biofilms that actually cause implant failure, analyzing biofilms directly attached to explanted implant surfaces may be more appropriate; however, such studies have been conducted very limitedly due to difficulties in sample acquisition. Meanwhile, recent studies have focused attention on the role of novel pathogens in peri-implantitis beyond traditionally known periodontal pathogens. Sulfate-reducing bacteria such as Desulfobulbus and Desulfovibrio can produce hydrogen sulfide, inducing tissue toxicity and inflammation, and associations with periodontitis and peri-implantitis have been reported in some studies (Cross et al., 2018). Additionally, Filifactor alocis, [Eubacterium] nodatum group, [Eubacterium] saphenum group, and Pseudoramibacter alactolyticus are emerging pathogens whose importance in periodontal disease has recently been highlighted, requiring further investigation of their role in peri-implantitis (Antezack et al., 2023; Lafaurie et al., 2023). Understanding the functional characteristics of microbial communities is also important for elucidating pathogenic mechanisms. Using functional prediction tools such as PICRUSt2 (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States), the metabolic potential of microbial communities can be inferred from 16S rRNA gene data. This approach allows identification of metabolic pathways activated in disease states beyond simple taxonomic compositional changes, which can contribute to discovering novel therapeutic targets. Therefore, this study aimed to compare microbial communities of biofilms collected from explanted implant surfaces due to peri-implantitis with subgingival plaque from healthy controls through 16S rRNA gene sequencing, and to identify differences in functional characteristics using PICRUSt2 (Douglas et al., 2020). Through this approach, we sought to expand understanding of microbiological factors involved in the etiology of peri-implantitis and to explore potential potential candidate and therapeutic targets.

This study analyzed the microbial community of biofilms collected directly from explanted implant surfaces due to peri-implantitis through 16S rRNA gene sequencing and compared them with subgingival plaque from healthy controls. The results revealed distinct differences in microbial community structure between the peri-implantitis and healthy control groups, confirming that sulfate-reducing bacteria and various emerging pathogens, in addition to well-known periodontal pathogens, are associated with peri-implantitis. Additionally, functional prediction analysis identified predicted metabolic pathway enrichment in peri-implantitis.

Comparison with Previous Studies.

The microbial profile associated with peri-implantitis identified in this study was largely consistent with previous research findings. The Red complex, consisting of Porphyromonas gingivalis, Tannerella forsythia, and Treponema denticola, has been reported as a core pathogen not only in periodontitis but also in peri-implantitis, and these bacteria were significantly increased in the peri-implantitis group in this study (Fernandes et al., 2024). Filifactor alocis has recently received attention as an important pathogen in periodontal disease, and in this study, it showed the highest LDA score in the peri-implantitis group, suggesting its potential as a strong biomarker (Antezack et al., 2023).

Health-associated bacteria predominant in the healthy control group, including Streptococcus, Neisseria, Rothia, Actinomyces, and Corynebacterium, are commensal bacteria commonly observed in healthy oral environments, consistent with previous studies (Anuntakarun et al., 2025; Bessa et al., 2025). These bacteria are known to contribute to maintaining oral homeostasis and play a role in inhibiting colonization by pathogenic bacteria.

The high R² value (0.596) observed in beta diversity analysis indicates that the difference in microbial community between peri-implantitis and healthy states is very large, which is similar to or higher than values reported in previous periodontitis studies (Joshi et al., 2025). These results support that peri-implantitis is a disease with distinct microbiological characteristics.

Novel Findings of This Study

This study presents novel findings that differentiate it from previous research in several aspects.

First, biofilms were collected and analyzed directly from explanted implant surfaces. Most previous studies collected samples from peri-implant pockets using paper points or curettes, which may reflect more planktonic bacteria within the pocket rather than biofilms directly attached to the implant surface (Faveri et al., 2014). The explanted implant surface biofilms used in this study have the advantage of more accurately reflecting the pathogenic community that actually caused implant failure.

Second, we confirmed that sulfate-reducing bacteria (SRB), Desulfobulbus and Desulfovibrio, were significantly associated with peri-implantitis. Sulfate-reducing bacteria can contribute to cytotoxicity, inflammation induction, and tissue destruction through the production of hydrogen sulfide (H₂S) (Kushkevych et al., 2020). Hydrogen sulfide is known to inhibit mitochondrial function, damage epithelial barriers, and promote inflammatory cytokine secretion. Although some studies have reported associations between periodontitis and sulfate-reducing bacteria, their role in peri-implantitis has not yet been sufficiently elucidated. Our findings suggest that sulfate-reducing bacteria may contribute to the etiology of peri-implantitis.

Third, emerging pathogens including [Eubacterium] nodatum group, [Eubacterium] saphenum group, Phocaeicola abscessus, Pseudoramibacter alactolyticus, and Pyramidobacter were significantly increased in the peri-implantitis group (Antezack et al., 2023). Although associations of these bacteria with periodontal disease have recently been reported, their role in peri-implantitis has been studied limitedly. Particularly, [Eubacterium] nodatum group and [Eubacterium] saphenum group are known to proliferate in anaerobic environments and possess proteolytic activity, potentially contributing to destruction of peri-implant tissues.

Fourth, predicted metabolic pathway enrichment in peri-implantitis were identified through functional prediction analysis using PICRUSt2. The pattern of increased pathways related to amino acid degradation, sulfur metabolism, iron acquisition, and LPS biosynthesis, along with decreased carbohydrate metabolism pathways in the peri-implantitis group, reflects a shift toward a proteolytic and anaerobic environment (Song et al., 2024). These functional changes may be closely associated with the establishment of a pathogenic environment in peri-implantitis and can contribute to discovering novel therapeutic targets.

Fifth, this study was conducted on Korean patients. Most previous peri-implantitis microbiome studies have been conducted on Western patients, and research on Asian patients, particularly Korean patients, is limited. Since the oral microbiome can differ according to genetic background, dietary habits, and lifestyle, our findings can contribute to understanding peri-implantitis microbiome characteristics in Korean patients (Kim et al., 2022).

Mechanistic Interpretation

The predicted enrichment of amino acid degradation pathways (glycine betaine degradation, L-glutamate degradation, L-lysine fermentation, etc.) in the peri- implantitis group suggests potential capacity for host tissue protein degradation. Metabolic products that could potentially be generated during these processes, including ammonia, hydrogen sulfide, and short-chain fatty acids (such as butyrate), might exhibit tissue toxicity and promote inflammatory responses. While PICRUSt2 predictions provide valuable hypotheses about community functional potential, these inferences should be validated through direct measurement of metabolic activity using metatranscriptomic or metametabolomic approaches (see Limitations) (Basic and Dahlén, 2023).

The increase in sulfur metabolism pathways is consistent with the increase in sulfate-reducing bacteria, and the hydrogen sulfide they produce can be involved in various pathogenic mechanisms including cytotoxicity, DNA damage, and inflammation induction. The increase in iron acquisition pathways reflects strategies for securing iron, which is essential for pathogen survival and proliferation, and can be interpreted as an adaptive mechanism of pathogens to counteract the host's iron restriction defense mechanism (nutritional immunity). The increase in LPS biosynthesis pathways is consistent with the increase in Gram-negative pathogens, and LPS potently activates the host inflammatory response through TLR4 receptors.

In contrast, the increased carbohydrate metabolism pathways in the healthy control group reflect a healthy oral environment where saccharolytic metabolism is predominant. The increase in sulfur assimilation pathways and aerobic degradation pathways indicates that aerobic or facultative anaerobic bacteria are predominant in healthy oral environments.

Clinical Implications

Our findings have several clinical implications that warrant careful consideration. First, the characteristic microbial profiles identified in peri- implantitis represent potential candidate biomarkers that require validation before clinical implementation. Particularly, Filifactor alocis, Desulfobulbus, and [Eubacterium] nodatum group are often not included in existing periodontal pathogen tests, suggesting the need to develop expanded microbial testing panels that include these bacteria. However, diagnostic utility must be established through: (1) validation in independent cohorts with adequate sample size, (2) assessment of diagnostic performance metrics (sensitivity, specificity, ROC analysis), (3) comparison with existing diagnostic methods, and (4) evaluation of clinical utility and cost-effectiveness. Our findings provide preliminary evidence supporting these bacteria as candidate markers, but prospective validation studies are essential before clinical translation (Antezack et al., 2023).

Second, with the confirmed role of sulfate-reducing bacteria and sulfur metabolism, the possibility of developing novel therapeutic strategies targeting hydrogen sulfide production inhibition or sulfate-reducing bacteria is suggested. Third, the activation of amino acid degradation and proteolytic pathways can provide a rationale for nutritional approaches that restrict protein substrate supply in peri-implantitis patients or for developing therapeutics that inhibit these metabolic pathways.

This study has several important limitations that must be carefully considered when interpreting our findings.

First and most critically, there is a fundamental difference in sample substrate between groups. The peri-implantitis group analyzed biofilms from titanium implant surfaces, while the healthy control group analyzed subgingival plaque from natural tooth surfaces. Titanium and tooth surfaces differ substantially in surface chemistry, energy, roughness, and electrical charge properties—all factors that strongly influence microbial adhesion and biofilm architecture independently of disease status. Additionally, peri-implant and periodontal tissues differ anatomically and immunologically, potentially creating different microenvironments that shape microbial communities. The observed microbial differences may therefore reflect: (1) disease-specific dysbiosis, (2) inherent surface-specific colonization patterns, or (3) a combination of both. We cannot definitively distinguish between these possibilities with our current study design. Ideally, biofilm from healthy implants would serve as the appropriate control, but such samples were not available due to ethical and practical challenges. Despite this limitation, our findings remain valuable because: (1) the identified pathogens align closely with previous peri-implantitis studies using pocket samples, (2) the strong clustering separation (R²=0.596) suggests robust disease associations, (3) our sensitivity analysis showed persistent dysbiosis independent of smoking (R²=0.177, p=0.001), and (4) predicted functional shifts align with known pathogenic mechanisms. Nevertheless, future studies using healthy implant biofilms as controls are essential for validation.

Second, substantial demographic differences exist between groups. The peri-implantitis group was older (62.3 vs 28.5 years) and predominantly male, while healthy controls were all female. Age, sex, and hormonal status influence oral microbiome composition, and these differences may contribute to observed variations. The demographic differences between groups (age, sex, smoking) are known to independently influence oral microbiome composition through multiple mechanisms. Aging is associated with decreased salivary flow, changes in immune function, and shifts toward more anaerobic communities. Sex hormones influence gingival vascularity and immune responses. Smoking creates a hypoxic environment favoring anaerobes and directly suppresses immune surveillance. When combined with different substrate properties (titanium vs tooth), these factors may interact synergistically to shape distinct microenvironments. While our sensitivity analysis suggests disease-associated dysbiosis persists independent of smoking, the reduced effect size (R²=0.177 vs 0.596) indicates that demographic factors substantially contribute to the observed variation.

Third, smoking status differed markedly. All healthy controls were non- smokers, while the peri-implantitis group included 9 smokers, 4 non- smokers, and 6 with uncertain status. Our sensitivity analysis comparing only non-smokers (n=4 peri-implantitis) showed persistent microbial differences, suggesting disease-associated dysbiosis exists independent of smoking, though the small sample size limits definitive conclusions.

Fourth, the retrospective biobank-based design precluded prospective demographic matching or multivariable adjustment for confounders.

Fifth, this cross-sectional study cannot establish causal relationships. We cannot distinguish whether microbial changes are a cause or consequence of peri-implantitis, nor can we track disease progression.

Sixth, PICRUSt2 infers functional potential from 16S rRNA sequences but does not measure actual gene expression, enzyme activity, or metabolite production. The predicted functional profiles represent potential metabolic capabilities rather than realized activities. Our findings should be interpreted as preliminary hypotheses requiring validation through metatranscriptomic or metametabolomic analysis. The NSTI values (mean < 0.15) indicate good prediction confidence, but direct validation remains essential.

Seventh, detailed implant characteristics (brand, surface topography) were unavailable due to biobank privacy policies. Implant surface heterogeneity may contribute to variability in microbial composition.

Future Research Directions

To address the limitations of this study, the following approaches are needed in future research. First, studies using control groups matched for age, sex, and smoking status are necessary. Second, longitudinal studies comparing microbial communities across stages of healthy implants, peri-implant mucositis, and peri-implantitis are needed. Third, basic research is needed to elucidate the pathogenic mechanisms of sulfate-reducing bacteria and emerging pathogens. Fourth, functional prediction results need to be validated through metatranscriptomic or metametabolomic analysis. Fifth, large-scale validation studies are needed to evaluate the diagnostic potential of the candidate biomarkers identified in this study. Specifically, future research should: (1) assess diagnostic performance (sensitivity, specificity, positive/negative predictive values) in independent cohorts, (2) determine optimal cutoff values for microbial abundance thresholds, (3) evaluate early diagnostic capability for detecting peri-implantitis before radiographic bone loss, (4) compare diagnostic accuracy with existing clinical and radiographic assessments, and (5) assess feasibility for point-of-care testing applications.

This study demonstrated that the microbial community of biofilms from explanted implant surfaces in peri-implantitis is distinctly different from healthy subgingival plaque, and that sulfate-reducing bacteria (Desulfobulbus, Desulfovibrio) and various emerging pathogens ([Eubacterium] nodatum group, [Eubacterium] saphenum group, Phocaeicola abscessus, Pseudoramibacter alactolyticus, etc.), in addition to known periodontal pathogens, are associated with peri- implantitis biofilms. Functional prediction analysis revealed predicted enrichment of pathways related to amino acid degradation, sulfur metabolism, iron acquisition, and LPS biosynthesis in peri-implantitis, along with predicted decreased carbohydrate metabolism. Despite the fundamental limitation of comparing biofilms from different substrates (titanium implants vs natural teeth), our findings showing strong microbial differentiation (R²=0.596) and persistent dysbiosis independent of smoking may contribute to expanding understanding of peri-implantitis etiology. The identified microbial profiles represent candidate biomarkers and potential therapeutic targets that warrant validation in independent cohorts using appropriate controls.