Soft tissue biopsies may prove culture negative in biofilm prosthetic infections. Identification of the causative bacteria could be achieved by either scraping of the prosthetic surface or by sonication of the entire implant. These techniques have not been thoroughly studied in experimental models where the biofilm is developed in vivo.

In a novel rat biofilm model we compared scraping and sonication as methods for dislodging biofilm bacteria.

Twenty plates of steel alloy (5×7×1mm), with a surface roughness (Ra) of 0.35 (0.19–0.51) μm, were inserted into 20 standardised pieces of sheep costae, weight: 1.2 (1.0–1.5) gram. To each bone graft was added 50 μL of a Staphylococcus epidermidis suspension containing 1.4 (1.1–1.7)×104 CFU. Ten Sprague Dawley rats were operated with implantation of the bone graft subfascially on each side of the interscapular region. After two weeks the grafts were excised. The plates were removed from the grafts and rinsed twice in saline. Aliquots of 50 μL were cultured. 10 plates were scraped, followed by vortex mixing of the knife blade; and 10 plates were sonicated at 30 kHz for five minutes. 50 μL of the saline used for a) vortex mixing of the knife blade, and b) sonication, was seeded on agar. After overnight incubation the number of CFU was counted.

The total number of CFU recovered after scraping and sonication were 2(0–13) × 102 and 298(8–878) × 102, respectively (p< 0, 01). Compared to the number of CFU in the rinsing fluid, no increase was observed after scraping. For each plate that was sonicated there was a 38 (3–300) fold increase in the number of CFU.

First, sonication is a superior technique for dislodging biofilm bacteria in an in vivo model, compared to scraping. Secondly, the present experimental model is a promising method for developing biofilm in vivo.

Culture of tissue samples obtained peri-operatively is ‘the gold standard’ for determining the presence of infection in prosthetic revision surgery. The growth of identical bacterial strains in three or more specimens strongly indicates an infected prosthesis. With routine microbiological culture techniques, identification of different phenotypes of coagulase negative staphylococcus (CNS) will be interpreted as either contamination or a polybacterial infection. At our clinic, different phenotypes of CNS are cultured in approximately 20% of patients operated with a two-stage revision due to a chronic prosthetic infection.

We studied the genotype of different phenotypes of CNS cultured in specimens obtained from prosthetic joints.

We analysed 22 cases, where different phenotypes of CNS were cultured in tissue, and joint fluid specimens were collected peri-operatively. The pre-operative diagnosis was chronic prosthetic infection (n=16), aseptic revision (n=5) and primary prosthesis (n=1). Different phenotypes were assessed by colony morphology and/or antibiogram. Pulsed-field gel electrophoresis (PFGE) was employed to identify and compare the genotypes.

In 16 out of 22 cases (73 %), PFGE unveiled that phenotypically different strains of CNS belonged to the same genotype. Of these 16 cases 7 had different antibiograms. In the other group (6/22), phenotypically different strains of CNS did not belong to the same genotype. In the 16 cases with different phenotypes belonging to the same genotype, gentamicin bone cement had previously been used in 15 cases. In the other group (6/22), gentamicin bone cement had not been used previously in any case (p < 0.01, chi-square test).

Phenotypically different strains of CNS identified by routine microbiological techniques should not be classified readily as contamination or as a mixed bacterial infection in prosthetic surgery. A particular precaution should be taken in the case of patients who had previously been operated on with use of gentamicin-loaded bone cement.

Low virulent chronic prosthetic infection might be indistinguishable to aseptic loosening. Polymerase chain reaction (PCR) has been introduced to improve bacterial detection of implant infection. However, there is great risk of false positive results when using broad range/universal PCR primers. The source of DNA contaminants may be of environmental origin as well as the reagents employed.

To study the presence of bacterial DNA in culture negative biopsy specimens by using defined criteria for PCR positivity.

We included six specimens from each of 21 patients preoperatively considered having aseptic loosening of a hip prosthesis. These 126 specimens were culture negative after seven days of incubation. Prior to incubation, the specimens were divided, and half of each specimen was processed for PCR.

Three sets of primer pairs targeting the 16S rRNA gene were developed.

Nine specimens from culture proven prosthetic infections and nine specimens from primary prosthetic surgery served as positive and negative controls, respectively.

A specimen was considered PCR positive if:

bacterial DNA ≥ 2 times than the reagent control, and

All the 126 patient specimens were PCR negative and the nine positive controls were positive. One of nine negative specimen controls was PCR positive, DNA sequencing demonstrated a non-pathogen. Five single PCR reactions (1.3 %) were positive.

This study underlines the importance of establishing stringent criteria for interpretation of PCR positivity to apply to clinical specimens. By the criteria used, five single positive PCR reactions were identified as false positives. Positive PCR reactions in all of the specimen positive controls prove the detection ability of the method. One PCR positive result in a specimen negative control demonstrates the contamination risk in collection and handling of biopsies.

In conclusion, by using a semiquantitative PCR and stringent criteria for PCR positivity we were unable to detect any infections missed by culture.