Aims: Identifying low-grade infection in failed total hip replacements (THR) is an important but often difficult task. Recently, there has been interest in the use of molecular biology techniques as potential sensitive tests for low-grade infection by identifying fragments of bacterial DNA within human tissue. Methods: We investigated the ability of a molecular biology technique known as the polymerase chain reaction (PCR) to identify low-grade infection during revision of THR considered to have failed from aseptic causes. We analysed 113 specimens of tissue and synovial fluid from 31 THR revised for aseptic loosening and compared them to 105 control specimens taken during 28 primary THR. All cases were performed in laminar flow theatres. No primary or revision specimen had positive microbiological cultures. No revision specimen had histological evidence suggestive of infection. Results: Using PCR, we identified bacterial DNA in 39 of 85 revision THR tissue specimens (46%) compared to 18 of 84 primary THR specimens (21.4%, p=0.001). Bacterial DNA was identified within the synovial fluid in three specimens taken from 28 revision THR (10.7%) and in two specimens taken from 21 primary THR (9.5%, p=0.36). As multiple specimens were sent per case, 16 of 31 revision THR (52%) and eight of 28 primary THR (29%) were considered to be infected (p=0.072). Conclusions: Our results suggest that many aseptically loose revision THR actually contain bacterial DNA within the peri-prosthetic tissue, but infrequently within the synovial fluid. With an overall specimen contamination rate of 19%, however, PCR has poor specificity for routine diagnostic use in revision THR.

Differentiating cases of aseptic loosening of total hip arthroplasty (THA) from loosening due to low-grade infection can often be difficult. It is possible that some cases of ‘aseptic’ loosening may be related to unidentified bacterial infection.

Using Polymerase Chain Reaction (PCR), this study attempted to identify the frequency with which bacterial DNA could be observed at revision arthroplasty for what was considered ‘aseptic’ loosening.

All revision cases had to fulfil strict criteria to be considered aseptically loose In all cases operative specimens from the synovial fluid, synovium, femoral and acetabular membranes where possible were sent for analysis by histology, bacteriology and by PCR to identify the presence of the 16S bacterial ribosomal fraction, an indicator of bacterial DNA. Ten bacteria per millilitre of tissue/fluid were the threshold for detection. As a control for environmental contamination, specimens from primary THA were also sent for analysis in the same manner as revisions.

The identification of bacterial DNA in at least one sample from a patient was considered a positive case result.

45 revision THA were identified over a 3-year period (1998–2001). From those 45 revision cases, 163 specimens were sent for analysis by PCR. These specimens were compared to the control group of 34 primary THA from which 91 specimens were sent for analysis by PCR. When analysed by specimens positive by PCR, bacterial DNA was identified in 55 of 163 specimens sent from the 45 revision THA. This compared with 21 of 91 specimens positive by PCR sent from the 34 primary THA (p=0. 07).

When analysed by cases positive by PCR, bacterial DNA was identified in 29 of 45 revision THA and in 8 of 34 primary THA (p< 0. 001).

PCR is a sensitive test for detecting infection in revision THA. Results from the primary THA cases would suggest there is at least a 23% false positive rate even with negative bacterial culture. The increased frequency with which bacterial DNA has been identified in ‘aseptically’ loose revision THAs, however, is unlikely to be due solely to environmental contamination. These results may have relevance for our interpretation and understanding of aseptic loosening as well for the diagnosis of prosthetic infection.