72 revision surgeries for aseptic failure after hip or knee arthroplasty: a prospective study with an extended diagnostic algorithm

The number of patients with failing total hip (THA) or knee (TKA) arthroplasties is rising due to the wide access to primary arthroplasties and the increasing longevity of patients [1, 2]. Scandinavian arthroplasty registries report aseptic failure (AF) as the most common indication for revision surgery followed by periprosthetic joint infection (PJI) [3, 4]. AF is a collective term for aseptic conditions, most of which are implant loosening, instability, and polyethylene wear [5].

It remains controversial whether some cases of AF may indeed be caused by low-grade infection associated with microorganisms forming biofilms [6‐8]. Biofilm-associated microorganisms are able to evade the immune response and thus typically produce a poor local and systemic host reaction [9].

At the present time, no gold standard exists for diagnosis of PJI. Pre-operative diagnostics include patient history, clinical examination, X-ray imaging, blood biochemistry, and synovial fluid analysis by microscopy and culture [10]. Intra-operatively macroscopic findings are essential and various samples can be obtained for definitive diagnosis. For several decades culturing of synovial tissue biopsies from the vicinity of the implant has been a mainstay for ruling out PJI [11‐13]. Several institutes define PJI as 2 positive periprosthetic cultures with phenotypically identical organisms [14, 15]. However no consensus exists. The concern for biofilm-forming bacteria has brought attention to the implant as the nidus of infection, and samples can be prepared in the laboratory by sonication of removed prosthesis parts [16]. New diagnostic options have been provided by modification of well-established culture methods and a range of DNA-based techniques.

This development has led to a number of studies critically examining the confirmation of AF [17‐21]. There may be reservations when interpreting such studies especially concerning the design, selection of patients, compliance with a strict protocol, standardization, and availability of novel techniques. Head-to-head comparisons of different methods have rarely been possible and their diagnostic accuracy remains debatable [22].

With an extensive diagnostic work-up, it may be possible with greater certainty to determine the prevalence of unrecognized PJI infection in patients undergoing revision for AF. The aim was to characterize AF patients with a range of supplementary microbiological diagnostic methods applied to standard and experimental specimen types obtained intraoperatively.

Patients were invited to participate in a study with the primary aim to improve the diagnosis of prosthesis-related infection and pain (Danish acronym PRIS) through a multidisciplinary diagnostic algorithm [23]. The project was carried out between 2011 and 2014 in the North Denmark Region, with a population of approximately 580.000.

During the study period, 156 patients were included. Seventy-two revisions were performed on the indication of AF in 71 patients (35 THA and 37 TKA; flow is depicted in Fig. 1. Data for the remaining 85 included on the suspicion of PJI or pain not shown). Patient characteristics are shown in Table 1. A re-classification was based on the laboratory work on the project samples: Confirmed AF, PJI and PJI-indeterminable status (Table 2).

At inclusion, 64 patients had pain as the dominant symptom followed by reduced mobility (Table 1). Four patients had an indolent prosthetic joint had joint swelling and reduced mobility. Two patients had a preoperative joint aspiration at the discretion of the surgeon. Culturing and molecular tests were negative in both cases. Six patients were scheduled for revision surgery following multi-modal nuclear imaging which showed no pathological uptake on dual white blood cell (WBC)/bone marrow scans in all the cases (Fig. 1). Fluor-deoxy-glucose (FDG) PET/CT scans showed FDG uptake in the periprosthetic synovial tissue or in the interface between bone and implant in five out of the six cases, compatible with both PJI and AF. Clinical acumen supported revision for AF.

In the present prospective study, we report a cohort of patients undergoing revision and investigated for AF according to a diagnostic algorithm and undergoing revision surgery [23]. We evaluated results for 71 patients undergoing revision surgery for AF after THA and TKA within a larger study addressing the diagnosis of prosthesis-related infection and pain. There is no gold standard for diagnosis of PJI, and therefore we maintained the culturing of five periprosthetic synovial tissue biopsies as the reference method. This has been the routine for approximately 20 years in our orthopaedic department [13], and on this background, we judged cautiously the results obtained with new sample types, modified culture methods, and molecular tests.

We found five of 72 cases with PJI not recognized prior to surgery. Three cases were detected with our reference standard, i.e. periprosthetic synovial tissues biopsies cultured for 6 days. Two additional cases were found by extended incubation (14 days) of joint fluid and cultivation of sonication fluid within the framework of the project. Combining standard and project diagnostics, complete negative results were obtained for 52 patients (53 procedures, 73.6%). Twelve patients had spurious positive findings on standard and project diagnostics deemed clinically insignificant (see Appendix 3).

WBC and CRP did not differ between the three groups. We do not consider the CRP difference between group 1 and 3 clinically relevant, since there are only three observations in group 3. Our findings with normal BMI (< 25) and not having positive periprosthetic synovial tissue biopsy cultures was in agreement with others [26].

We obtained concordant results for cultures of synovial tissue biopsies and sonication fluid in three PJI patients. Results discordant with synovial tissue biopsies were found in three patients. One was re-classified PJI-indeterminable (group 1 case 11). The two last patients were re-classified to confirmed AF (group 2 case 7 and group 1 case 14).

Relatively few studies of patients with a failing THA or TKA have focused entirely on AF and applied a broad range of laboratory diagnostic methods [18, 20]. Ribera et al. [19] studied 89 patients with AF as defined by clinical and radiological criteria and absence of local signs or symptoms of infection. PJI was defined by histopathology, intraoperative purulence, and at least two cultures with the same pathogen. Disregarding results from sonication fluid, 13% (12/89) were diagnosed with occult PJI. Concordance between cultures from tissue samples and sonication fluid was 75% (9/12).

Trampuz et al. [27] analyzed cultures of synovial tissue biopsies and sonication fluid in 331 patients undergoing revision for PJI or AF [27]. AF (n = 252) was a postoperative diagnosis defined by prosthetic failure and absence of PJI criteria which mainly were overt signs of infection and positive histopathology. Twenty-one patients with a diagnosis of AF had a single positive tissue culture and two had two or more positive cultures. Culturing of sonication fluid was positive in three cases (1.2%) although tissue culture was negative. Direct comparison with our study is difficult, but unmistakably positive cultures seemed to be rare in patients undergoing revision for AF.

Fernandez-Sampedro et al. [20] compared cultures from sonication fluid and periprosthetic tissue biopsies in patients undergoing revision for AF (n = 198). Pre- and intraoperative diagnoses were based on strict criteria including radiological and radionuclide imaging, the absence of clinical signs of infection, and normal inflammatory markers. Similar to Trampuz et al., postoperative diagnoses of AF were established by the absence of PJI. Furthermore, isolation of a microorganism from sonication fluid in an amount of at least 20 colony forming units or a single positive culture of S. aureus or S. lugdunensis were a part of the definition. Based on these criteria a diagnosis of AF was made in 174 cases (88%). Disregarding histopathology, 16 patients (8.1%) had occult PJI based on positive cultures from periprosthetic tissue and sonication fluid. Positive culture of sonication fluid, but negative cultures of periprosthetic tissue were found in 5 cases (2.5%). Compared with our study the frequency of PJI in Spanish patients undergoing revision for AF was notably higher than in our study (12, 95% CI: 8–17%). Previous antibiotic therapy may mask PJI at least for a period of time, and it is noteworthy that few Danish patients had been treated incidentally for another reason. The Spanish study does not provide this information, but it is noteworthy that antibiotics are available without prescription in Spain [28]. Antibiotics are a prescription drug in Denmark.

Periprosthetic tissue and project samples were examined with 16S rRNA gene PCR and amplicon sequencing in our study. Three of five patients re-classified to PJI had confirmatory PCR findings for sonication fluid. Of three patients previously treated with antibiotics, none had positive findings by PCR in the collective set of standard and protocol samples. Gomez et al. [29] evaluated 16S rRNA gene PCR performed with sonication fluid in 231 patients with AF diagnosed postoperatively similar to Trampuz et al. Three patients had positive PCR, but negative sonication fluid and tissue culture. Two of these patients had prior antibiotic treatment. Information on the indication for revision and re-classification was not reported. Molecular methods have theoretical advantages for detecting microbial pathogens in association with prosthetic failure. Currently, this attracts much interest, and our findings suggest that experienced surgeons can distinguish PJI and AF from each other with high accuracy and we found little support for the hypothesis that occult infection is a frequent cause of AF.

Results from the different studies including this performed under the PRIS project did not support a significant value of routine molecular tests of intraoperative tissue in patients suspected of AF. Details from 16S rRNA in this study have been published elsewhere [23].

A major strength of our study was the algorithm rooted in a clinical setting. The strict protocol of intraoperative tissue collection and the post-operative laboratory analysis led to an unbiased comparison of culture-dependent and independent methods. We feel confident that missing data from inclusion on patient history and during follow-up was controlled by linkage to the medical records, laboratory and microbiological database. All patients are registered in the hospital and regional administrative data system allowing for merging and linking to the clinical databases. Additionally a close collaboration between the participating clinical specialties was valuable for patient management.

Our study has a number of limitations. The inclusion of patients followed a pragmatic study design. Distinguishing between AF and PJI was done by a senior Orthopaedic clinician and followed the existing practice in our Institute. This included patient history, physical examination and X-ray. Thus the pre-operative identification of AF, including the six patients with previous multimodal nuclear imaging, was predominantly a clinical decision. These issues reflect the challenges in this field.

Furthermore the low number of patients and consequently few infected cases complicates conclusion. Histopathology and biochemical analysis of synovial fluid were not part of the algorithm and are not used in Denmark on a routine basis. This hampered comparisons with other European studies. Biochemical analyses were not specified as inclusion criteria, and CRP values were incomplete in 13 patients. Nevertheless, three of five patients with unrecognized PJI had elevated CRP values. These three patients were not excluded, as it was not deemed clinically relevant at inclusion. Several studies of the diagnostic utility of inflammatory markers have been published recently [30, 31]. Biochemistry including CRP should be part of pre-operative evaluation, however it cannot stand alone. In other studies it is reported as a minor criteria for PJI [14]. FDG PET/CT scans was not a planned diagnostic method for AF, nevertheless clinical acumen supported revision in 6 cases.

In conclusion, our results showed a low incidence of PJI in cases suspected of AF pre-operatively in patients with THA or TKA. Experimental specimen types and prolonged cultures were conducive to detection of two of five PJI cases unrecognized prior to surgery. We were able to control for missed PJI diagnoses through access to health databases within the North Denmark Region. One patient presented with PJI deemed unrelated to the failure of the previous arthroplasty.

A structured approach is advisable to identify unrecognized PJI in revision for AF. The algorithm served as a useful tool. However, clinical judgment should not be outweighed in the pre-operative decision-making process. Extended incubation improved diagnostics in accordance with the existing literature. Sonication of retrieved components did not provide additional information in this study. A previous paper with a methodological aim analyzed in depth the differential contribution of specimen types, extended incubation of cultures, and 16S rRNA sequencing from this study.