The present clinical study investigated 53 dental implants from 49 patients. Main focus of the study was a microbial comparison of implants which got lost during the time of osseointegration (early loss) and implant loss because of peri-implantitis (late loss). All results were collated by the same dentist at the Dental Academy for Continuing Professional Development (Karlsruhe, Germany).
General inclusion criteria were: implants with severe loss of the bone-to-implant contact, either in the period of healing after the insertion (< 3 months, “early loss”) or after uneventful osseointegration and loading (> 3 months, “late loss”). Exclusion criteria were: patients aged younger than 18 years, radiation or bisphosphonate therapy, patients with untreated periodontitis and severe general conditions, such as uncontrolled diabetes, tumors, severe heart disease or a reduced state of general health. Smokers were not excluded from the study. Included were only patients with one or more implants which had been inserted in our clinic. Only implants with severe peri-implant bone loss and no chance of preservation were included. Also specimens from implants without peri-implantitis (no bone loss) were integrated into the study as a control group.
Implants with peri-implantitis and bone loss, but chances of preservation were excluded.
Study population
Patients were divided into four groups:
27 patients with severe peri-implantitis and implants without a chance of preservation were divided in two groups.
- Group E: early implant loss (implants with severe peri-implantitis during osseointegration prior to prosthetic restoration, ≤ 3 months after implant placement), 13 patients with 14 implants
- Group L: late implant loss (implants with severe peri-implantitis and prosthetic restoration for more than three years), 14 patients with 17 implants
For a comparison to the healthy oral situation we created two control groups:
- Group CE: control group (implants with no bone loss directly after completed osseointegration, two to four months after implant placement), 17 patients with 17 implants
- Group CL: control group (implants with no bone loss and prosthetic restoration for more than three years), 5 patients with 5 implants
The observational study was approved by the ethical review committee of the local medical association (Institutional Review Board of the Saarland Medical Council, Germany; ID: 232/12) and was conducted in accordance with the Declaration of Helsinki and the Professional Code for Physicians of the local Medical Council. All patients were informed about the purpose of the study by the examiner and signed a form of consent.
Clinical procedure of documentation of clinical findings and sampling
- Patients were asked about their case history, nicotine abuse, diabetes, regular use of mouth rinses (at least once per week) and antibiotics in the past 12 months, and if periodontitis had been diagnosed.
- In each patient a pool sample was taken with sterile paper points from the peri-implant sulcus around infected or healthy implants. (Fig. 1). Any existing signs of inflammation (pocket suppuration) were documented.
- If a severe peri-implantitis was diagnosed, explantation was done under local anaesthesia with articaine with epinephrine 1:100.000 (Citocartin Sopira®, Heraeus Kulzer GmbH, Hanau, Germany). The removed implants themselves were also used for microbiological analysis for validation of the paper point results.
All patients were informed in advance about the clinical procedure to be performed. In addition, the patients were informed about obtaining paper point samples for a future microbial analysis in connection with the study. All patients gave their informed consent in writing.
For the microbial analysis of the peri-implant tissue the bacterial DNA from the paper point samples of a total of 53 implants from 49 patients were obtained and analyzed for taxonomic composition by sequencing the V1-V2 variable regions of the 16S rRNA gene. The paper points were applicated for 20 seconds, following the established clinical routine for sample collection of peri-implant pathogens.
Evaluation at patient level
For the analysis at patient level, every implant was weighted inversely to the total number of implants per patient.
Extraction of microbial DNA
Sterile paper points were used to collect biofilm samples of the peri-implant sulcus for microbiota analysis as described previously(23). Briefly, biofilm microbes were resuspended in nuclease free water using a combination of shaking and sonication. The suspension was centrifuged, pellets were stored at -80°C and then used for DNA extraction using commercial extraction protocols for genomic DNA (QIAamp Mini Kit, Qiagen, Hilden, Germany). The collected pellets of the supernatant were treated with 180 µL lysozyme solution (20 mg/mL, SIGMA-Aldrich, Taufkirchen, Germany; 20 mM Tris-HCl, pH 8.0, 2 mM EDTA, 1.2% Triton) under shaking at 37°C for 2:15 h, followed by proteinase K digestion (20 µL proteinase K and 200 µL buffer AL) for 1:15 h under shaking at 56°C. Finally, the DNA was eluted with 100 µL PCR-clean water and the concentration was quantified using the NanoDrop equipment (PEQLAB, Erlangen, Germany). One empty extraction without any sample material was used a control for background DNA contaminations (contamination control).
Illumina sequencing of amplicons targeting the 16S rRNA gene
Amplicons of the V1-V2 region of the bacterial 16S rRNA gene were prepared as published elsewhere (24). Briefly, the genomic sequence of the 16S rRNA gene was amplified with primers that were derived from the previously described primers 27F and 338R (25, 26) and contained sequences compatible with Illumina sequencing platforms and a 6-nt barcode sequence. The resulting DNA was used as template for a second PCR, using primers designed to introduce full-length Illumina adapter sequences including Illumina 6-nt index sequences to enable high-level multiplexing. To control for potential DNA contamination an additional sample was amplified without template DNA (contamination control sample). Libraries were pooled and subjected to 250 nt paired-end sequencing on an Illumina MiSeq machine. For data analysis, the obtained reads were processed using dada2 version 1.16.0 (27) following the MiSeq SOP published by Kozich et al. (28). Dada2 determines amplified sequence variants (ASVs) by removing sequencing errors based, which tends to produce more accurate results than the commonly used operational taxonomic units or OTUs (29). Briefly, sequence reads were trimmed to a length of 200 nt (forward read) and 150 nt (reverse read) and 5 nt of the left end were trimmed additionally. Reads with ambiguous base calls and an expected error rate larger than 2 were discarded and ASVs were inferred using the pseudo-pooling algorithm. Chimeric sequences were removed and ASVs classified using the RDP Naïve Bayesian Classifier algorithm as implemented in dada2 against the Silva database v138 (30). Were possible by exact matching, ASVs were assigned species names using the assignSpecies function of dada2. After the preprocessing the data amounted to between 1,305 and 205,073 sequences per sample.
Statistical analysis of microbiome data
Further analysis of the microbiome data was done in R version 3.6.1 (31) using packages phyloseq (32), vegan (33) and ggplot2 (34). The contamination control sample yielded 397 reads and 13 ASVs represented by more than 5 sequences in the contamination control were removed from all samples. Alpha-diversity of the sampled biofilms was estimated using the Inverse Simpson Index, calculated by the phyloseq function estimate_richness. Principal coordinates analyses (PCoA) were performed using the ordinate function with Jenson-Shannon Divergence (JSD) as distance metric. Differences of single variables between groups of samples were statistically tested using the Kruskal-Wallis Rank Sum Test (R function kruskal.test) for multiple groups and Wilcoxon Rank Sum Test (R function pairwise.wilcox.test) for pairwise comparisons of groups with the Benjamini-Hochberg adjustment to control false discovery rate in multiple comparisons (35). Differences in the microbiota composition (beta-diversity) were tested by performing Permutational Multivariate Analysis of Variance (PERMANOVA) on the JSD matrix, using the adonis function of the R package vegan (33) and pairwise Adonis by Martinez Arbizu & Monteux (36).
The study was conducted in accordance with the STROBE guidelines.