Comparative analysis of the microbiota in gingival crevicular fluid from peri-implantitis and periodontitis using 16 S ribosomal RNA gene amplicon sequencing: a cross-sectional study Authors: Hongyu Liu, Jiawen Chen, Qiong Chen, Lingling Zhang, Yiqi Feng, Buling Wu, Jingyi Wu Published: 2025 ABSTRACT AbstractObjectiveThis study aimed to investigate the diversity, composition, and functional characteristics of the gingival crevicular fluid (GCF) microbiota in individuals with periodontitis, peri-implantitis, and healthy controls (Groups P, PI, and H, respectively), and to identify potential biomarkers specific to PI.Materials and methodsGCF samples were collected from nine participants diagnosed with periodontitis, peri-implantitis, or classified as healthy. Microbial communities were analyzed using 16 S ribosomal RNA (rRNA) gene amplicon sequencing to profile the plaque microbiota.ResultsGroups P and PI showed significantly higher species richness than Group H. Principal coordinates analysis revealed that the microbial composition of Group PI was more similar to Group P than Group H. Firmicutes were enriched in Group PI at the phylum level, whereas Neisseria, Veillonella, Streptococcus, Haemophilus, and Prevotella were dominant at the genus level. Linear discriminant analysis effect size analysis identified Proteobacteria, Prevotella, Prevotella_7, Actinobacillus, and Lachnoanaerobaculum as significantly enriched in Group PI. Functional prediction showed associations with energy metabolism, cell wall synthesis, and amino acid synthesis. BugBase analysis revealed distinct phenotypic microbiota profiles in Group PI. Spearman correlation analysis indicated strong positive associations between PI-enriched genera and periodontitis-associated pathogens such as Fusobacterium.ConclusionThe microbiota of Group PI exhibits unique taxonomic and functional profiles. Prevotella and other taxa may serve as potential candidate biomarkers associated with PI. Their diagnostic efficacy and clinical applicability require further clarification through large-sample, multi-center validation studies. Functional enrichment in energy metabolism, cell wall synthesis, and amino acid synthesis pathways suggests a role in disease progression.Keywords: Peri-implantitis, Periodontitis, Gingival crevicular fluid, 16S ribosomal RNA, Microbiota Introduction Peri-implantitis is an inflammatory disease of the hard and soft tissues surrounding dental implants, characterized by progressive bone loss and representing a major challenge in dental implantation [1]. With an average prevalence of 19.53%, it is a leading cause of implant failure [2]. Clinically, PI is complex to manage and often requires palliative treatment, placing a significant burden on patients and society. It is fundamentally a chronic inflammatory process triggered by an imbalance between bacterial infection and host immune response. Although its plaque biofilm-mediated inflammation and progressive tissue destruction resemble periodontitis, key differences exist in microbial ecology and host response [3–5]. Unlike natural teeth, implants lack a periodontal ligament, and bacteria colonize the rough implant surface. This difference contributes to the often more rapid progression of PI compared with periodontitis, reflecting distinct microbial colonization and immune interactions [6–9]. Periodontitis is a chronic inflammatory disease driven by microbiota, characterized by the red complex of bacteria—including Porphyromonas gingivalis, Porphyromonas tartarica dense spirochetes, and Fusobacterium fusiformis—which produce virulence factors that destroy periodontal tissue [10]. Globally, periodontitis affects 20%–50% of the population [11–13], leading not only to tooth loss but also to an increased risk of systemic diseases [14–18]. Studies have shown significant differences between the pathogenic microbial communities of periodontitis and PI, indicating that the pathogenic mechanisms and therapeutic targets of PI cannot be directly inferred from those of periodontitis. Recent high-throughput sequencing reveals a unique microbial niche at the implant–tissue interface, with two key features: a marked increase in non-classical periodontal pathogens (e.g., Treponema, Prevotella, Filifactor) [19–21], and high microbial community heterogeneity [22, 23]. These features likely reflect the combined influence of physicochemical properties of the implant surface, the local microenvironment, and host immune responses, suggesting that PI develops through pathogenic mechanisms distinct from those of periodontitis. In recent years, next-generation sequencing technology [24] has significantly advanced the analysis of complex microbial ecosystems associated with oral diseases [25–27]. Among next-generation sequencing methods, 16 S ribosomal RNA gene amplicon sequencing has become central for analyzing microbial community composition, structure, and function, offering high-resolution taxonomic and quantitative analyses [28, 29]. Unlike traditional culture-based methods, it overcomes the limitations of cultivating fastidious or low-abundance species, enabling efficient detection and significantly broadening the scope of microbial research. Numerous studies have investigated microbial diversity in periodontitis, confirming the presence of distinct microorganisms and complex community shifts in affected individuals. However, the microbial analysis of PI remains incomplete. Current research gaps include: (1) an unclear relationship between the microbial ecology of PI and periodontitis; (2) limited identification of specific microorganisms as diagnostic biomarkers for PI; and (3) insufficient understanding of the roles of dominant bacterial genera in disease progression. Notably, existing investigations often focus on subgingival plaque as the primary sample type, which primarily reflects the “community structure” of microbial colonization but fails to capture the dynamic metabolic interactions between microbes and the host [20]. In contrast, the GCF, which contains not only microbes but also their metabolic products, host inflammatory factors, and tissue exudates, offers a more direct window into the real-time interplay of “microbial functional activity” and disease progression. This study leverages GCF as a key sample type to address these gaps, distinguishing it from plaque-centric prior work. Therefore, this study aims to comprehensively analyze the microbial communities in patients with PI using 16 S rRNA gene amplicon sequencing. By comparing findings from individuals with periodontitis and healthy controls, the study seeks to define PI-specific microbial features, clarify its relationship with periodontitis, identify diagnostic biomarkers, and assess links between species diversity and disease progression—ultimately providing a theoretical basis for the precise diagnosis and treatment of PI. Materials and methods Subjects and clinical examinations This study was approved by the Ethics Committee of the Stomatological Hospital of Southern Medical University (Protocol Number: 2023-58), and written informed consent was obtained from all participants. GCF samples were collected from patients randomly recruited during routine diagnosis and treatment in the Department of Implantology. According to the definition of periodontal health in the 2017 World Workshop on the Classification of Periodontal and Peri-Implant Diseases and Conditions [30], it covers two scenarios: one is clinical health on an intact periodontium, and the other is clinical gingival health on a reduced periodontium (including stable periodontitis patients and non-periodontitis patients). Healthy volunteers included in this study were those with clinical health on an intact periodontium, specifically: periodontal pocket depth (PPD) ≤ 3 mm, no clinical attachment level (CAL) loss or radiological bone loss, and bleeding on probing (BOP) at < 10% of sites. The inclusion criteria were as follows: (1) at least 20 remaining teeth; (2) diagnosis of PI or periodontitis based on the 2017 World Workshop classification guidelines [31]; and (3) good general health and willingness to cooperate. The exclusion criteria were as follows: (1) pregnancy or breastfeeding; (2) acute oral infections or chronic mucosal lesions; (3) harmful habits such as heavy smoking (> 20 cigarettes per day) or excessive betel nut chewing (> 5 betel nuts per day); (4) systemic diseases affecting periodontal status; (5) periodontal treatment within the past three months; and (6) use of systemic antibiotics or anti-inflammatory drugs within the past six months. A total of nine subjects were recruited (three each in the healthy, PI, and periodontitis groups; hereafter referred to as Group H, Group PI, and Group P, respectively), all in good general health. Oral clinical examinations were performed on selected teeth and implants in each subject. Specifically, GCF samples were collected from 1 tooth or 1 implant per subject. Examined teeth or implants meet the inclusion criteria of their respective groups. For subjects in Group P and Group PI, the selected teeth or implants are required to be free from the acute inflammatory phase and best represent the subject’s disease status. Sample collection GCF samples were collected from around eligible teeth or implants in the nine recruited subjects. The gingiva at each sampling site was dried with a sterile cotton ball, placed in the interdental space or oral vestibule to isolate saliva. A sterile filter paper strip was gently inserted into the gingival sulcus (2–3 mm depth) for 1–2 min, then removed and placed in a sterile centrifuge tube containing 1 mL of sterile distilled water. If GCF volume was insufficient, a micropipette was used to aspirate and transfer it into the tube [32]. Samples were mixed and centrifuged at 12,000 × g for 5 min. Extraction of total DNA Total microbial DNA was extracted using the cetyltrimethylammonium bromide method from microbiome samples of various sources. DNA quality was assessed by agarose gel electrophoresis, and quantity was measured using an ultraviolet spectrophotometer. Extracted DNA was stored at − 80 °C for future use. Polymerase chain reaction (PCR) amplification and purification The V3–V4 region of the bacterial 16 S rRNA gene was amplified using primers 341 F (5’-CCTACGGGGNGGCWGCAG-3’) and 805R (5’-GACTACHVGGGGTATCTAATCC-3’) through PCR. The amplification protocol was: 98 °C for 30 s; 35 cycles of 98 °C for 10 s, 54 °C for 30 s, and 72 °C for 45 s; followed by 72 °C for 10 min. PCR products were verified by 2% agarose gel electrophoresis, purified with AMPure XT beads, and quantified using a Qubit fluorometer. Purified products were evaluated with an Agilent 2100 Bioanalyzer and Illumina library quantification kit. Qualified libraries were serially diluted, pooled, denatured, and sequenced on a NovaSeq 6000 platform using paired-end 2 × 250 bp sequencing [33]. Bioinformatics analysis Raw sequencing data were assembled, quality-controlled, and filtered for chimeras to generate high-quality CleanData. Length filtering and denoising were performed using the qiime dada2 denoise-paired command via DADA2, yielding the amplicon sequence variant feature table and representative sequences. Subsequent analyses included diversity analysis, taxonomic annotation, and differential analysis. Alpha diversity was evaluated using the Chao1 and Shannon indices, whereas beta diversity was assessed using Jaccard distances and visualized via principal coordinates analysis. Differential abundance analysis was performed based on species abundance profiles. The Mann-Whitney U test or t test was used for comparisons between two groups, and the Kruskal–Wallis test for multiple groups. A p-value less than 0.05 was considered statistically significant. Results The clinical characteristics of subjects in Groups H, P and PI were listed in Table 1. Compared with Group H, the mean values of PPD and gingival index (GI) in Groups P and PI were significantly different (p < 0.01). In addition, the mean value of PPD, CAL and GI in Group PI was significantly higher than that in Group P (p < 0.05). Table 1Characteristics of the subjectsParametersH (n = 3)P (n = 3)PI (n = 3)P value - H-PP value - H-PIP value - P-PIPatient age (years)34.67 ± 8.3363 ± 11.5358 ± 17.06---Sex - Male1 (33.3)1 (33.3)2 (66.7)---Sex - Female2 (66.7)2 (66.7)1 (33.3)---PPD (mm; average of all sites)2.38 ± 0.084.72 ± 0.895.36 ± 0.94< 0.01< 0.01< 0.05PPD (mm; subgingival sampled site)N/A6.13 ± 0.236.97 ± 0.21N/AN/A< 0.01CAL (mm)N/A3.13 ± 0.233.97 ± 0.21N/AN/A< 0.01GI0.08 ± 0.141.58 ± 0.511.83 ± 0.39< 0.01< 0.01< 0.05PPD periodontal pocket depth, CAL clinical attachment level, GI gingival index, PH periodontally healthy, PT periodontitis, PI peri-implantitis, N/A not availableValues are presented as mean ± standard deviation or number (%). P < 0.05, analysis of variance with the post hoc Games-Howell test Characteristics of the subjects PPD periodontal pocket depth, CAL clinical attachment level, GI gingival index, PH periodontally healthy, PT periodontitis, PI peri-implantitis, N/A not available Values are presented as mean ± standard deviation or number (%). P < 0.05, analysis of variance with the post hoc Games-Howell test Microbial community diversity The Chao1 index indicated greater species richness in Groups P and PI compared with Group H, although the differences were not statistically significant (Fig. 1A). The Shannon index revealed a higher trend in Groups P and PI compared with Group H, which indicated a higher community diversity trend, though the differences were not statistically significant (Fig. 1B). As shown in Fig. 1C, samples within Groups H, P, and PI clustered closely, tentatively suggesting consistent microbial community structures within each group. To further assess beta diversity, permutational multivariate analysis of variance was performed. The analysis revealed significant differences in microbial community structures among the groups, helping to mitigate bias from the small sample size. Specifically, samples in Group H were clearly separated from those in the diseased groups (Groups P and PI), whereas Groups P and PI showed partial overlap. This may indicate potential distinctions in microbial community profiles between healthy and diseased states, with some similarities, but also potential discernible differences, between Groups P and PI. The alpha and Beta diversities of the microbial communities in the P, PI, and H groups. (A-B) The alpha diversity values in the GCF samples were calculated by the CHAO1 index and the Shannon index. (Colors denote groups: red for periodontitis (P), blue for peri-implantitis (PI), and green for periodontally healthy (H)) (C) Beta diversity is a PCoA plot generated using Jaccard distance. (Colors denote groups: red for periodontally healthy (H), blue for peri-implantitis (PI), and green for periodontitis (P)) Relative abundance of microbial composition At the phylum level (Fig. 2A), the six most abundant phyla in GCF samples were Proteobacteria, Firmicutes, Bacteroidetes, Fusobacteria, Actinobacteria, and Spirochaetes. The diseased groups tended to show higher relative abundances of Firmicutes and Bacteroidetes than Group H, whereas Group H appeared to have more Actinobacteria. The relative abundance of Proteobacteria was similar in Groups H and PI, both generally higher than in Group P. Between the diseased groups, Group PI exhibited a tendency toward higher levels of Proteobacteria and Firmicutes, but relatively lower levels of Bacteroidetes, Fusobacteria, Actinobacteria, and Spirochaetes compared with Group P. Heat map analysis (Fig. 2B) revealed that microbial compositions at the phylum level in Groups P and PI were partially similar, but showed clear differences from Group H. Moreover, compared with Group H, the diseased groups tended to exhibit increased relative abundances of more microbial taxa. Composition characteristics and differences of microbial communities in GCF samples from the P, PI, and H groups at the phylum and genus taxonomic levels. A-B Analysis of microbiota composition at the phylum level. (C-D) Analysis of microbiota composition at the genus level. (P: periodontitis, PI: peri-implantitis, H: periodontally healthy. In Figs. A and C, the x-axis shows sample groups, the y-axis represents the relative abundance of a taxon, and different colors correspond to different species at the same level. In Figs. B and D, each row denotes a species, each column a sample group, with a blue-to-red gradient indicating increasing abundance.) At the genus level, 12 genera—Neisseria, Porphyromonas, Fusobacterium, Veillonella, Streptococcus, Haemophilus, Rothia, Acinetobacter, Capnocytophaga, Leptotrichia, Prevotella, and Treponema—together accounted for over 50% of total sequences in each group (Groups H, P, and PI; Fig. 2C). Group H tended to show higher relative abundances of Rothia, Acinetobacter, Capnocytophaga, Leptotrichia, Aggregatibacter, Lautropia, and Actinomyces compared with the diseased groups. In Group PI, Neisseria, Veillonella, Streptococcus, Haemophilus, and Prevotella were relatively more abundant than in Groups H and P. Conversely, Porphyromonas, Fusobacterium, and Treponema were relatively more abundant in Group P than in Group PI. Heat map analysis at the genus level (Fig. 2D) showed partial similarity between Groups P and PI, but both were clearly distinct from Group H, suggesting that diseased states may be associated with the enrichment or depletion of specific genera. Comparison of significantly different relative abundances Linear discriminant analysis effect size analysis (Fig. 3A and C) revealed significant differences in the relative abundances of multiple species between the diseased groups and Group H (p < 0.05), with 32 species differing in Group P and 42 in Group PI. Notably, the relative abundances of Bacteroidales, Peptostreptococcales, Filifactor, Synergistes, Fretibacterium, Synergistaceae, Synergistales, Synergistota, Lachnospirales, and Lachnospiraceae exhibited a significant increasing trend in both Groups P and PI (p < 0.05). A higher absolute linear discriminant analysis score indicated a greater difference in species abundance between groups, which may suggest a stronger discriminatory role and potential as a candidate diagnostic biomarker. As shown in Fig. 3A, Spirochaetae, Treponema, Spirochaetaceae, and Clostridia were differentially enriched with a notable trend in Group P, whereas Fig. 3C shows that Prevotellaceae, Prevotella, and Prevotella_7 were differentially enriched with a notable trend in Group PI. These taxa may potentially serve as candidate biomarkers for periodontitis or PI. In Fig. 3B and D, red nodes indicate species enriched in Group H, whereas green nodes indicate those enriched in the diseased groups. Significantly different relative abundances of microbiota according to the disease status. A-B The relative abundances of the microbiota in the H group and the P group are significantly different (Colors denote groups: red for periodontally healthy (H) and green for periodontitis (P)). (C-D) The relative abundances of the microbiota in the H group and the PI group are significantly different (Colors denote groups: red for periodontally healthy (H) and green for peri-implantitis (PI)). (Figs. A and C show significantly different species with an LDA score > 3.0 (preset), i.e., statistically different biomarkers. In Figs. B and D, circular layers radiating from inside out represent seven taxonomic ranks. (kingdom, phylum, class, order, family, genus, species)) When comparing the diseased groups, 10 species showed a significantly higher trend in Group PI (p < 0.05; Fig. 4A), including Proteobacteria, Negativicutes, Veillonellaceae, Veillonellales, Prevotella_7, Actinobacillus, and Lachnoanaerobaculum. In contrast, 22 species were found to have a significantly higher trend in abundant in Group P than in Group PI (p < 0.05; Fig. 4A), such as Clostridia_vadinBB60, Sphingomonas_sp_ASBCFS01, Catonella, Tannerella, Lentimicrobium, Stenotrophomonas, and Treponema. In Fig. 4B, red nodes indicate enrichment in Group P, whereas green nodes indicate enrichment in Group PI. Significantly different relative abundances of microbiota between subjects with P and subjects with PI (Colors denote groups: red for periodontitis (P) and green for peri-implantitis (PI)). (Figs. A show significantly different species with an LDA score > 3.0 (preset), i.e., statistically different biomarkers. In Figs. B, circular layers radiating from inside out represent seven taxonomic ranks. (kingdom, phylum, class, order, family, genus, species)) Relative abundances of significant bacterial genus and species Genus-level differential analysis is shown in Fig. 5A. Compared with Groups H and PI, Group P exhibited a significantly higher trend in the relative abundances of Filifactor, Fretibacterium, Lentimicrobium, Catonella, Rikenellaceae_RC9_gut, Clostridia_vadinBB60, Eubacterium__brachy, and Mycoplasma (p < 0.05). In contrast, Actinobacillus showed a significantly higher abundance trend in Group PI than in Groups H and P (p < 0.05). Species-level differential analysis is presented in Fig. 5B. In Group P, unclassified species within the above genera exhibited a significantly higher abundance trend than in Groups H and PI (p < 0.05). Conversely, unclassified species within Actinobacillus and Streptococcus showed a significantly higher abundance trend in Group PI than in Groups H and P (p < 0.05). These taxa may potentially serve as candidate key indicators for diagnosing periodontitis and PI, respectively. Furthermore, compared with Group H, the diseased groups showed a significantly reduced trend in the abundances of Acinetobacter sp. DSP1.1, Sphingomonas sp. JSM_221518, Acinetobacter sp. A14-49, and unclassified species within Sphingomonas and Paracoccus (p < 0.05), suggesting that these species may play important roles in maintaining a healthy oral microbial environment. Differential analysis of all species at the genus and species levels among the P, PI, and H groups. (A) Differential analysis of all species at the genus levels among the P, PI, and H groups. (B) Differential analysis of all species at the species levels among the P, PI, and H groups.(Colors denote groups: red for periodontally healthy(H), blue for periodontitis(P), and green for peri-implantitis(PI). The abscissa represents the differential species, and the ordinate represents the relative abundance.) Bacterial correlation networks Figure 6A shows pairwise correlations among the top 30 dominant genera in GCF samples from Groups H, P, and PI, with correlation coefficients |rho| > 0.8. Pathogenic genera associated with periodontitis, such as Fusobacterium and Treponema, were found to be significantly positively correlated with newly identified genera including Prevotella and Fretibacterium (p < 0.05). Prevotella, which had a significantly higher relative abundance in Group PI, also exhibited strong positive correlations with Fusobacterium, and Bergeyella was positively correlated with Capnocytophaga (p < 0.05). These genera have been reported to be involved in periodontitis, which may suggest potential synergistic interactions between bacterial communities in periodontitis and PI. In the network, Bergeyella and Capnocytophaga appeared to display high connectivity with other genera. Similarly, Fig. 6B presents genus-level correlations in heat map format for Groups H, P, and PI: blue squares indicate positive correlations and red squares indicate negative correlations. The closer the absolute value is to 1, the stronger the correlation. Positive correlations may imply synergistic coexistence, whereas negative correlations may suggest mutual exclusivity. The correlations between pairwise dominant microbial communities in the GCF samples of the subjects in the P, PI, and H group. (A) Pairwise correlations among the top 30 dominant genera in GCF samples from Groups H, P and PI. (B) Heatmap of genus-level correlations among Groups H, P and PI. (P: periodontitis, PI: peri-implantitis, H: periodontally healthy. In the network graph, different nodes represent distinct dominant genera; node colors indicate their phyla. Line thickness reflects correlation strength: solid lines for positive correlations, dashed lines for negative ones. Node size denotes the number of associated bacteria.) Bacterial function prediction Using PICRUSt2 for functional prediction, we performed annotations based on the Clusters of Orthologous Groups of proteins (COG) databases, focusing on the top 30 functions with p < 0.05 (Fig. 7A and C). Differences were observed between Group H and the diseased groups in multiple physiological processes, which may provide preliminary insights into disease mechanisms and potential candidate biomarkers (Fig. 7A and B). Compared with Group H, the COG-based annotation results showed that the microbiota in Group PI was found to be significantly associated with physiological processes related to substance metabolism and transformation, such as sulfatase maturase, phosphodiesterase, phosphoglycerol transferase, flavodoxin, and alcohol dehydrogenase YqhD (p < 0.05). In contrast, when compared with Group H, Group P was found to be significantly associated with physiological processes related to substance transport, energy metabolism, and oxidative stress repair, with enrichment in functions like ATPase, sodium-phosphate cotransporter, sodium or hydrogen transporting membrane pyrophosphatase, flavoreductase, and DNA repair photolyase (p < 0.05). This may be associated with the higher energy demand and oxidative stress level in the context of this disease state. Figure 7C revealed that compared with the Group P, the Group PI was mainly found to be significantly associated with physiological processes such as substance metabolism, transcriptional regulation, and lipid synthesis (p < 0.05), including functions of NADPH reductase, GreA/GreB family transcription elongation factors, and oxX_SURF4 family membrane proteins. While Group P was found to be significantly associated with physiological processes like energy supply, stress resistance regulation, and DNA repair (p < 0.05), such as ATPase domain, anti-σ regulatory factor, and DNA repair photolyase functions. In addition, pathway enrichment analysis (Fig. 7D and E) showed that compared with Group H, Group P was mainly observed to be enriched in pathways related to energy acquisition, amino acid metabolism, and substance synthesis, while the PI group was observed to be enriched in energy metabolism pathways. Figure 7F indicated that compared with Group P, Group PI was mainly observed to be enriched in cell wall synthesis and amino acid synthesis pathways, such as peptidoglycan maturation and L-arginine biosynthesis IV pathways, whereas the Group P was mainly observed to be enriched in bacterial cell wall synthesis-related metabolism and glycogen biosynthesis I pathway. These differences may suggest the potential of constructing a candidate oral microbiota metabolic network model. Future studies could explore potential possible cross-interactions using multi-omics technologies. These results may suggest that despite being in a disease state, the GCF microbiota in the PI group and P group exhibit distinct metabolic patterns. The COG databases were used to perform functional annotation on the differentially abundant microbiota in each group (A-C) and conduct pathway enrichment analysis (D-F). (Colors denote groups: blue for periodontitis (P), purple for peri-implantitis (PI), and orange for periodontally health (H)) Bacterial phenotype prediction BugBase, a microbiome phenotype prediction tool, was used to analyze the potential phenotypes of the GCF microbiota in Group H and the diseased groups. The results, presented in Fig. 8A and I, correspond to nine potential phenotypes, including Aerobic, Anaerobic, Contains Mobile Elements, Facultatively Anaerobic, Forms Biofilms, Gram Negative, Gram Positive, Potentially Pathogenic, and Stress Tolerant. The results showed that compared with Group H, the diseased groups exhibited an increasing trend in the relative abundance of microbiota with specific phenotypes at the phylum level. Among these, the microbial phenotypes with an increasing trend in Group PI relative to Group H included Aerobic, Contains Mobile Elements, Facultatively Anaerobic, Forms Biofilms, Gram Positive, Potentially Pathogenic, and Stress Tolerant. In contrast, the microbial phenotypes with an increasing trend in Group P relative to Group H included Aerobic, Anaerobic, Facultatively Anaerobic, Forms Biofilms, and Gram Negative. Compared with Group P, Group PI had higher relative abundances of microbiota with the phenotypes of Aerobic, Contains Mobile Elements, Facultatively Anaerobic, Forms Biofilms, Gram Positive, Potentially Pathogenic, and Stress Tolerant. These phenotype-based differences may serve as candidate biomarkers for the early diagnosis of periodontitis and peri-implantitis. Monitoring changes in the abundance of phenotype-associated microbiota may help assess disease progression. Furthermore, given the high relative abundance of microbiota with the biofilm-forming phenotype in the diseased groups, targeted therapies that disrupt or inhibit biofilms may improve the efficacy of antibacterial treatments. Probiotics can also be used to rebalance the oral microbiota, suppress harmful bacteria, and support beneficial ones, which may help enhance the treatment and prevention of oral diseases. Phenotypic prediction of the microbiota in the GCF of the P, PI and H group A-I. (The x-axis represents group names, and the y-axis shows the relative abundances of phylum-level species predicted to have different phenotypes. Different species can be distinguished by color, while the height of the bars indicates the relative abundances of different species across various groups. P: periodontitis, PI: peri-implantitis, H: periodontally healthy.) Discussion and conclusions This study used 16 S rRNA gene amplicon sequencing to systematically compare the GCF microbial communities across Groups H, P, and PI. At the phylum level, Firmicutes showed significantly higher relative abundance in Group PI. At the genus level, Neisseria, Veillonella, Haemophilus, and Prevotella were characteristically enriched in Group PI. PICRUSt2 functional prediction suggested that the microbiota in Group PI may be significantly enriched in energy metabolism, cell wall synthesis, and amino acid synthesis pathways. BugBase predictions identified differences in the relative abundances of specific phenotypic profiles between Groups PI and P. Spearman correlation analysis revealed significant positive correlations between pathogenic genera in Group P (e.g., Fusobacterium and Treponema) and enriched genera in Group PI (e.g., Prevotella). These predictive findings require further validation through functional experiments. The dynamic evolution of microbial community diversity and structure may play a role in the transition of the oral microecology from a healthy to a diseased state, providing preliminary insights into the pathogenic mechanisms of periodontitis and PI. Previous studies have shown that an inflammatory microenvironment can shift the microbiota from stable symbiosis to pathogenic commensalism, accompanied by increased alpha diversity and restructuring of key genera [20, 34–36]. Although there was no statistically significant difference in alpha diversity in this study, the trend was consistent with previous studies, indicating a higher trend in Groups P and PI compared with Group H. Although both diseased groups shared pro-inflammatory microbial features, they showed distinct separation at the genus and species levels. Beta diversity analysis revealed group-specific microbiota structures in GCF, with some overlap between Groups P and PI. Notably, prior studies have reported similar community separation between P and PI using subgingival or buccal plaque samples [20, 37], but their focus on microbial taxonomy alone limits insights into functional dynamics. In contrast, our analysis of GCF, by capturing both microbial taxa and their metabolic context, suggests that the observed overlap between P and PI is not merely structural but may reflect shared functional pathways alongside distinct genus-specific activities. This functional dimension, unique to GCF-based analysis, provides a basis for deepening our understanding of why PI and P share certain clinical features yet differ in progression and treatment response. Prior studies have similarly reported clear separation of the microbiota between Groups P and PI in subgingival and buccal plaque samples [20, 37]. The present findings highlight the potential diagnostic value of GCF in distinguishing healthy and diseased states and confirm statistically significant intergroup differences in beta diversity through permutational multivariate analysis of variance, reducing the potential bias from the small sample size. However, it is still necessary to expand the sample size for further verification to increase the reliability of the conclusion. In terms of microbial composition, the increased relative abundance of Firmicutes in both Groups P and PI may represent a shared pro-inflammatory feature of these diseases. In this study, Firmicutes and Bacteroidetes were more abundant in the diseased groups than in Group H. This is consistent with previous research reports that have emphasized the dominant status of Firmicutes in peri-implantitis and its close association with inflammation [38, 39]. At the genus level, the microbial profiles of Group P and PI exhibited both overlaps and differences. Porphyromonas, Fusobacterium, and Treponema were more enriched in Group P, whereas Group PI was characterized by higher abundances of Neisseria, Veillonella, Streptococcus, Haemophilus, and Prevotella. Although some of these genera, such as Streptococcus and Prevotella, have previously been confirmed to be associated with peri-implantitis [37, 38, 40, 41], the present study identified, through GCF analysis, that Neisseria, Veillonella, and Haemophilus were significantly more enriched in Group PI compared with Group P. This discrepancy may stem from the ability of GCF to capture microbes that actively interact with host tissues, given that these genera are known to secrete host-interactive factors.These factors may not be representative in dental plaque which mainly reflects the structure of adhesive biofilm. Recent studies have further revealed the key roles and potential mechanisms of Neisseria, Veillonella, and Haemophilus in the oral inflammatory microenvironment, providing important insights into the significance of their enrichment in PI. Among these genera, Neisseria and Haemophilus are significantly enriched in the lesional tissues of patients with oral lichen planus [42], suggesting their potential involvement in regulating the dysregulation of the oral immune microenvironment. Existing literature has also confirmed that Neisseria can secrete Immunoglobulin A protease [43], which impairs the host mucosal immune barrier, this explains the potential mechanism by which its enrichment in the PI group may exacerbate local immune dysregulation around implants. Furthermore, as a core bridging genus in oral biofilm formation, Veillonella can support the growth of periodontal pathogens and play a crucial role in dental biofilm formation and periodontitis progression [44]. Meanwhile, its surface protein Hag1 can provide adhesion sites for Porphyromonas gingivalis [45], indirectly influencing the occurrence and development of PI. These findings support the reliability of our data and suggest that these genera may be closely associated with PI. Follow-up studies should expand the sample size, include multicenter cohorts, and employ metagenomic sequencing to validate the stability of these differential genera across clinical settings. At the species level, linear discriminant analysis effect size analysis revealed significantly increased relative abundances of 12 species, including Filifactor, Synergistes, and Synergistota, in both Groups P and PI. Moreover, 10 species, including Actinobacillus, Prevotella, Prevotella_7, and Lachnoanaerobaculum, were significantly more abundant in Group PI than in Group P. Differential analysis at both genus and species levels also showed that Group PI had unique taxa with significantly higher relative abundances compared with Groups H and P. A previous study reported significant increases in Anaerovibrio fermentans, Bacteroides vulgatus, Faecalibacterium prausnitzii, Olsenella umbonata, Parasutterella hominis, Prevotella buccalis, Pseudoramibacter alactolyticus, Treponema microdentium, and Slackia exigua in Group PI20. These elevated genera and species under inflammatory conditions may represent key pathogens or pathogenic symbionts that could contribute to the progression of PI. However, the relationship between these taxa and clinical indicators, such as probing depth and bone resorption, remains to be clarified. Longitudinal studies are needed to quantify associations between microbial markers and disease severity and should be validated through clinical and experimental methods. Differences between this study and others may stem from variations in sample sources, sizes, selection criteria, detection techniques, analytical methods, disease stages, or duration. These findings suggest that although periodontitis and PI share certain microbial features, PI is not merely “periodontitis around an implant,” but may instead represent a distinct oral disease with a unique microbiota-driven mechanism. Synergistic interactions within microbial communities represent a potential critical entry point for understanding the pathogenic mechanisms of periodontitis and PI. Microbial network construction and analysis serve as key tools to reveal such interactions. Previous studies have shown that Treponema, a major pathogenic genus in periodontitis, may also play a central role in PI [46]. This study further revealed that periodontitis-associated pathogens such as Fusobacterium and Treponema were closely linked to Prevotella and Fretibacterium, both significantly enriched in Group PI. Furthermore, Prevotella and Bergeyella, which showed higher relative abundances in PI, were positively correlated with Fusobacterium and Capnocytophaga, respectively. These associations suggest that periodontitis-related pathogenic genera may contribute to PI through synergistic pathogenic effects, supporting previous findings that patients with a history of periodontitis have an elevated risk of developing PI21. Unlike plaque-focused studies that emphasize co-colonization patterns, our GCF-based correlation analysis highlights potential functional synergy. For example, the link between Prevotella and Fusobacterium may reflect metabolic cross-feeding within the host-derived fluid environment, which is less pronounced in adherent plaque. This functional interaction, captured uniquely in GCF, provides a tentative mechanistic explanation for why periodontitis history increases PI risk. Mechanistically, this study found that Prevotella was significantly associated with the expression of NOD-like receptor family pyrin domain-containing 3 and absent in melanoma 2 inflammasomes, along with their effectors caspase-1 and interleukin-1β, suggesting a potential role in the progression of PI via inflammasome pathway activation [3]. Network analysis also showed that Bergeyella and Capnocytophaga had high connectivity with other taxa, implying that they may act as “hub genera” that regulate microbial community structure and promote stability of pathogenic consortia through metabolic interactions or niche competition. Collectively, these findings suggest that key genera may cooperate via shared metabolic or signaling pathways, potentially influencing periodontal tissue health. However, the molecular mechanisms underlying these interactions remain unclear. Future research should integrate metagenomics, single-cell sequencing, and in vitro/in vivo studies to clarify inter-genus interactions and their impact on host immunity,, which could facilitate precise targeting of pathogenic microbial networks. Clinically, enhanced monitoring for PI may be warranted in patients with a history of periodontitis. Microbial regulation strategies targeting hub genera warrant further exploration as potential approaches for the combined prevention and treatment of periodontitis and PI. Understanding the pathogenic mechanisms of periodontitis and PI requires examining the functional metabolic characteristics and phenotypic differences of microbial communities, which together may reflect disease heterogeneity. From a metabolic perspective, imbalance in substance and energy metabolism appears to be a potential shared pathological feature of both diseases, while the specific enrichment of amino acid synthesis and glycogen biosynthesis pathways suggests the existence of distinct metabolic drivers. In this study, functional prediction using PICRUSt2 linked the microbiota in Groups P and PI to metabolism-related functions. This focus on functional pathways represents a key advance over prior taxonomic studies. GCF’s unique content of microbial enzymes and host-derived metabolites allowed us to identify these pathways, which may be critical for microbial survival in the implant microenvironment. Such functional insights, inaccessible through plaque analysis alone, lay a foundation for exploring PI-specific therapeutic targets. Prior studies had shown that pathway alterations in GCF samples from patients with periodontitis primarily involved cell motility, translation, cell growth and death, and metabolism, but systematic functional analyses of PI-related microbiota remain limited [47]. These findings suggest that substance and energy metabolism may play a key role in the pathogenesis of both diseases. Moreover, distinct metabolic signatures were identified: the cell wall synthesis and amino acid synthesis pathways were uniquely enriched in Group PI, whereas the enrichment of bacterial cell wall synthesis-related metabolism and glycogen biosynthesis I pathways were enriched in Group P. The former may potentially disrupt the implant-bone integration through “enhanced colonization and autonomous nutrient synthesis”, whereas the latter could potentially maintain chronic inflammation and tissue destruction through “optimized adhesion and energy storage”. Further experimental validation is required. Regarding phenotypic differences, variations in the abundance of microbiota with specific traits in Groups P and PI may reflect the selective pressure exerted by local microenvironments. In particular, biofilm-forming phenotypes may serve as potential key indicators of disease progression and treatment response. This study used BugBase for phenotypic prediction and identified differences between periodontitis and PI, potentially linked to enhanced biofilm formation promoted by the hydrophilicity of implant surfaces and the emergence of acid-producing phenotypes driven by inflammation in periodontal pockets. However, current literature lacks comparative analyses of site-specific microbial phenotypes in PI and their correlations with clinical indicators. Future studies should integrate metagenomic, metabolomic, and phenomic data to construct a multidimensional regulatory model that links metabolism-related pathway enrichment with phenotypic evolution. Emphasis should be placed on exploring how metabolism-dependent pathways may influence key genera through cell wall synthesis and amino acid synthesis processes. Machine learning algorithms can be applied to investigate the associations between functional modules and phenotypic traits in large-scale databases. Furthermore, multicenter cohort studies are required to validate the diagnostic value of metabolism-related markers and microbial phenotypes in PI, which could provide translational evidence to guide the development of metabolism-regulating coating materials and the formulation of phenotype-targeted therapeutic strategies [48–51]. This study identified potential unique compositional, functional, and phenotypic features of PI-associated microbiota using 16 S rRNA gene amplicon sequencing, providing multidimensional preliminary evidence for understanding the pathogenesis of PI and guiding clinical intervention. Compared with traditional subgingival plaque samples, this study focused on GCF samples for analysis. It not only identified the significant enrichment of Firmicutes in the GCF of PI patients but also clarified the enrichment characteristics of specific genera such as Neisseria, Prevotella, and Veillonella. This finding not only supports the existence of a shared basis of microbial dysregulation between PI and periodontitis but also highlights the structural differences in the microbiota shaped by the implant microenvironment. Meanwhile, it provides “targeted candidate potential biomarkers” for subsequent large-sample validation studies, effectively avoiding the “blindness” often associated with large-sample research. In addition, these findings may offer some new perspectives on the traditional view that “PI is simply periodontitis around implants,” while also providing microbiomic evidence to support the notion that PI is different from periodontitis. Clinically, the confirmed positive correlation between a history of periodontitis and increased risk of PI supports the potential need for a microbiota-based monitoring and early warning system for patients with periodontitis undergoing implant restoration. Although this study is limited by sample size and technical constraints, it nevertheless lays a preliminary foundation for constructing a precision medicine model that connects microbial markers, clinical phenotypes, and treatment response. This study has several limitations, which are detailed as follows: First, in terms of study design, this is a cross-sectional study. Such a design cannot establish a temporal causal relationship between microbial changes and disease progression; it can only reflect the correlation between microbial community characteristics and disease status at a specific time point, making it difficult to trace the dynamic interaction process between the two. Second, this study is limited by a small sample size, a single-center cohort, and insufficient consideration of host factors, which may restrict the representativeness and generalizability of the research results. Third, there are obvious limitations at the technical method level, the 16 S rRNA gene amplicon sequencing technology used in this study can only provide microbial taxonomic information and functional inference results based on prediction algorithms but cannot directly obtain empirical evidence of microbial functions. Conclusions regarding functional associations therefore require further verification through more precise multi-omics technologies such as metagenomics in subsequent studies. Additionally, there were oversights in the experimental quality control process. This study did not set up extraction blanks or PCR negative controls, making it impossible to effectively exclude exogenous contamination that might be introduced during sample extraction and amplification. This poses a potential risk to the accuracy of the microbial community analysis results. To address the above limitations, future research can be improved in the following aspects: First, expand the sample size and extend to multi-center cohorts, including research subjects from different regions and with different clinical characteristics to enhance the generalizability of the results. Second, incorporate long-term dynamic monitoring and use a longitudinal study design to track the temporal relationship between the microbiota and disease progression, providing a basis for causal inference. Third, integrate multi-omics research methods and combine them with in vitro functional experiments to explore the functions of specific microorganisms and their mechanisms of action in diseases more deeply. Fourth, strictly implement quality control measures, such as adding extraction blanks, PCR negative controls, and positive controls, to comprehensively control the risk of experimental contamination. Fifth, strengthen the comprehensive evaluation of host variables and verify the clinical value of microbial markers through multi-center clinical trials. These efforts will help strengthen the theoretical foundation and provide practical guidance for the early diagnosis, precise treatment, and effective prevention of PI. In summary, the core value of this study lies in attempting to advance oral microecology research from describing microbiota differences to exploring underlying mechanisms, functions, and phenotypes. The identified PI-specific microbial features and metabolic networks offer potential candidates for early diagnostic markers and novel antibacterial materials. Moreover, they support the possibility of a paradigm shift in oral implantology, from mechanical repair to ecological regulation, providing a preliminary clinically translatable research model for the precision prevention and management of PI. Acknowledgements This work was supported by grants from the National Natural Science Foundation of China (81900956), the President Foundation of Southern Medical University Shenzhen Stomatology Hospital (Pingshan) (2022A001, 2024A001), the Science and Technology Program of Guangzhou (2025A04J3459) and the Scientific Research Training Program of Stomatological Hospital of Southern Medical University (PY2023022、PY2023018). Abbreviations Gingival crevicular fluid Periodontitis, peri-implantitis, and healthy controls Ribosomal RNA Periodontal pocket depth Clinical attachment level Bleeding on probing Polymerase chain reaction Gingival index Clusters of Orthologous Groups Authors’ contributions Hongyu Liu: Methodology; data curation; investigation; validation; formal analysis; writing-original draft; writing-review and editing. Jiawen Chen: Methodology; data curation; investigation. Qiong Chen: Conceptualization; methodology; visualization. Lingling Zhang: Methodology; data curation; investigation. Yiqi Feng: Methodology; software; formal analysis. Buling Wu: Writing-review and editing; funding acquisition; project administration; resources. Jingyi Wu: Writing-review and editing; funding acquisition; project administration; resources; supervision. Funding This work was supported by grants from the National Natural Science Foundation of China (81900956), the President Foundation of Southern Medical University Shenzhen Stomatology Hospital (Pingshan) (2022A001, 2024A001), the Science and Technology Program of Guangzhou (2025A04J3459) and the Scientific Research Training Program of Stomatological Hospital of Southern Medical University (PY2023022、PY2023018). Data availability The datasets generated and/or analyzed during the current study are available in the Genome Sequence Archive (GSA) repository, with the accession number (BioProject) PRJCA045154 (https://ngdc.cncb.ac.cn/search/specific?db=bioproject&q=PRJCA045154). The data that support the findings of this study are available on request from the corresponding author. Declarations Ethics approval and consent to participate This study design was permitted by the Ethics Committee of Stomatological Hospital of Southern Medical University, Guangdong Province, China (Prot. Number 2023-58). Written informed consent was obtained from all participants, and the study was conducted in accordance with the Declaration of Helsinki revised in 2008. All patients were thoroughly informed about the research and completed an informed consent form. Consent for publication All authors consent to the publication of this manuscript. Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Buling Wu, Email: bulingwu@smu.edu.cn. Jingyi Wu, Email: Jingyiwu1121@163.com. References Associated Data Data Availability Statement The datasets generated and/or analyzed during the current study are available in the Genome Sequence Archive (GSA) repository, with the accession number (BioProject) PRJCA045154 (https://ngdc.cncb.ac.cn/search/specific?db=bioproject&q=PRJCA045154). The data that support the findings of this study are available on request from the corresponding author.