The State of Microbiology Diagnostic of Prosthetic Joint Infection in Europe: An In-Depth Survey Among Clinical Microbiologists

Background This study aims to give an overview on how microbiology diagnosis tests of Prosthetic joint infections (PJI) is performed in Europe, and to explore whether any factor influences the decision on implementing a test. Methods An extensive online survey of clinical microbiologists from seven European countries (Belgium, Estonia, Germany, Italy, Netherlands, Switzerland, and Spain). Following items were assessed: (i). general information on the laboratory, (ii) preference of the laboratory and clinical microbiologists regarding samples, (iii) transportation and (iv) processing of explanted foreign bodies and tissues and synovial fluid, (v) culture media and culture duration, (vi) reporting (identification and susceptibility testing), and (vii) use of molecular microbiology techniques. Results Invited were 163 clinical microbiologists. The response rate from each country was above 50% (range 51–78%), except for Germany (36%). Frequent PJI diagnostics were the use of tissue pre-processing (58.1%), culturing synovial fluid in blood culture bottles (45.5%), use of sonication for processing explanted prosthesis (56.8%), reporting the presence of synovial leukocyte counts (67%), use of blood aerobic and anaerobic agar (97.7%), and enrichment media thioglycolate (69.3%). The most common incubation time of the culture media is 7–14 days (34.1–70.5%). The clinicians were called to report the culture results (80.7%), and to give antibiotic recommendation (67%). Conclusion There are common practices in processing PJI samples and reporting results, which is promising for harmonization of PJI diagnostic in the future. However, variation in diagnostic tests should also be considered in interpreting and comparing clinical microbiology results.


INTRODUCTION
Prosthetic joint infections (PJI) occur in 0.5-2% of the performed arthroplasties (Tande and Patel, 2014). Due to the aging population, the number of joint arthroplasty is predicted to be increased (Ackerman et al., 2019;Matsuoka et al., 2021). Consequently, it will increase the numbers of PJI. PJI is managed by a combination of antimicrobial therapy and surgery (Yusuf and Borens, 2015). The diagnostic procedure for PJI should be highly reliable because the invasive nature, and long duration of required treatment. To oversee the diagnosis, the role of clinical microbiology laboratories and clinical microbiologists is paramount.
There are several diagnostic microbiology approaches that are applicable mostly for PJI diagnosis and less in other types of infections, such as obtaining multiple tissue samples (Atkins et al., 1998) and releasing bacteria from the removed joint prosthesis by sonication (Trampuz et al., 2007), vortexing (Portillo et al., 2013), or addition of dithiothreitol (Sambri et al., 2018). Several novel but even less standardized laboratory tests in PJI diagnostic have also been introduced, such as microcalorimetry of synovial fluid (Morgenstern et al., 2020), and the application of minimal biofilm inhibitory concentration (MBIC) or minimum biofilm eradication concentration (MBEC) (Sandoe et al., 2006). Since their original publications, the diagnostic methods are continually evaluated, and the conclusion is not always unequivocal (Dudareva et al., 2018;Yan et al., 2018). Most clinical microbiology laboratories offer a selection of tests based on guidelines, recommendation, and own experience. The tests performed may thus vary.
Identifying the heterogeneity of diagnostic approaches for PJI is needed to understand variable results and consequently variation in the PJI prevalence. The lack of standardization of methods needs to be assessed too. The aim of this in-depth survey is to give an overview of on how microbiology diagnosis is performed in several European countries, and to explore which factor influences the decision on implementing a test.

Survey Strategy
The survey was conducted among clinical microbiologists in seven European countries (Belgium, Estonia, Germany, Italy, Netherlands, Switzerland, and Spain) in English language. Survey questions were developed by two clinical microbiologists (EY and TK). In September 2020, the survey was piloted among seven clinical microbiologists from the participating countries. The survey was validated by asking their co-workers to fill in the survey. The validation showed >95% concordance of the responses to the survey questions. These clinical microbiologists were further requested to prepare a list of potential clinical microbiologists participants for this survey. The potential participants should be representative of clinical microbiologists in their countries (i.e., working in various geographic areas and from academic and non-academic hospitals). The potential clinical microbiologist to be invited can only represent a single clinical microbiology laboratory. The invitation to participate was sent to the e-mails of possible participants using a personal link in June 2021. It was allowed to refer the invitation to a different colleague, technician or trainee in that laboratory. Perhaps due to the corona pandemic, the response rate in July 2021 was as low as 30% after three reminders within 3 weeks sent by e-mail. Therefore, it was decided to change the approach for nonresponders, by sending them a personal email. The survey was closed 6 month after initial invitation.

Survey Domains
This was an in-depth survey asking laboratory protocols in details. It was estimated that 30 min was needed to answer the questions. There were 32 questions divided into 7 sections (Supplementary File 1): general information regarding the laboratory and number of PJI diagnosis, preference of the laboratory and clinical microbiologists regarding the samples, transporting the samples (explanted foreign bodies and tissues), processing the samples (tissues, explanted foreign bodies, and synovial fluid), culture media and culture duration, reporting (identification and susceptibility testing), and molecular microbiology techniques.

Statistical Analysis
Frequencies and percentages of variables were presented. Spearman and binary logistic regression analysis were performed to explore the correlation and association between the 2 variables, respectively. All analyses were performed using IBM SPSS Statistics for Windows, Version 26.0 (IBM Corp., Armonk, NY).

Samples and Transportation
Most (n = 71/88, 80.7%) of the laboratories had a protocol on sampling. The majority of clinical laboratories (n = 79/88, 89.8%) reported that the samples they received were mostly tissue biopsies ( Table 2).
Using logistic regression analysis, we found an association between the frequency of swabs of peri-operative tissue submitted to the laboratories with the distance between the hospital and the laboratory. Laboratory with the distance >10 km from the hospital had an odds ratio (OR) of 3.0 (95% CI 1.01-8.9) of frequently submitting swabs of operative tissue than laboratory situated in the same building as the hospital.
A total of 64 laboratories (72.7%) reported that they received tissue materials mostly in dry container, 34 (38.6%) mostly in saline solution, 9 in enrichment media (10.2%), and 3 (3.4%) in Ringer's solution. Among the laboratories that received foreign body samples (n = 69, 78.4%), 55 reported that they received the material mostly in dry container, 20 mostly in saline solution, 5 in Ringer's solution, 5 in broth, and 1 in Micro DTTtect (total count exceeds 100% because multiple answers were allowed).

Processing Tissue Samples
Almost all labs (n = 86, 97.7%) had a protocol for processing tissue samples from patients with suspected PJI (Table 3).

Processing Explanted Foreign Bodies
The participating laboratories that processed explanted foreign body materials and performed sonication (n = 50/88, 56.8%), added Ringer's or saline solution (n = 37/50, 74.0%). Most of these laboratories (n = 36/50, 72%) did not recommend pre-determined volume, but it was based on the size of the prosthetic and the container. Standardized recommended pre-determined volume used sonication procedure varied, up to 500 ml. The resulting fluid after sonication underwent centrifugation according to 24/50 (48%) of the laboratories processing these type of samples. Only a minority of the laboratories prescribed that the resulting sonication fluid should be inoculated to blood culture bottles (n = 17/50, 34%) and mostly they prescribed 10 ml (n = 10) to be inoculated in these bottles.
The major part of the laboratories called the clinicians to report the culture results of presumed PJI (n = 71/88, 80.7%), and 59 (67%) also gave antibiotic recommendation. Around 3 quarter (n = 67/88, 76.1%) of the laboratories had an access to the clinical data of the patients, and 49 (55.7%) also participated in multidisciplinary meetings with clinicians. The larger the distance between the laboratory and hospital, the lower the odds ratio to have access to clinical data and the larger the size of the hospitals, the larger the odds ratio to have PJI multidisciplinary meetings ( Table 4).

Molecular Microbiology Methods
Less than half (n = 37/88, 42%) of the laboratories had the capacity to perform molecular microbiology tests. Most of them (n = 34/37, 91.8%) could perform 16s RNA, and 6 used commercial multiplex PCR tests (BioFire R , Unyvero, GeneXpert R ). None of the participating laboratories used whole genome sequence in diagnosing PJI. The PCR test was never performed solely without cultures. Only two laboratories always performed the PCR test along with cultures. Most of the laboratories performed PCR upon request (n = 27/88, 30.7%) or when the culture was negative (n = 19/88, 21.6%).

DISCUSSION
There are several important findings from the survey. As expected, the microbiology procedures vary among the participating laboratories. However, there are several procedures that can be considered as typical clinical microbiology tests in diagnosing PJI, such as culture of the tissues and the prosthesis explants, where >50% of the participating laboratories processed these materials using tissue processing and sonication, respectively. Culture of synovial fluid and leukocyte count in it is also common. Mostparticipants incubated the materials for less than 14 days. Semi quantitative reporting was the most common method and the clinical microbiology results are phoned to the clinicians. Other common procedures are the use of MALDI-TOF for identification, automated antimicrobial susceptibility test, and the use of EUCAST breakpoints. Clinical microbiology criteria from organizations, such as Musculoskeletal Infection Society (MSIS) , Infectious Diseases Society of America (IDSA) (Osmon et al., 2013), and European Bone and Joint Infection Society (EBJIS) (McNally et al., 2021), overlap ( Table 5), but there are fine distinctions. MSIS and IDSA advised the requirement of at least 3 periprosthetic intraoperative tissue samples, but the maximum differs. MSIS asked the maximum of 5, while IDSA advised 5 or 6 without a maximum. Since tissues are obtained by invasive methods, efforts should be taken to maximize microbiological yield and to prevent contamination. The simplest procedure is by streaking and rolling tissues on solid culture media. However, this method does not expose the inside parts of the tissue to the surface of the media, and it may reduce the detection rate of the microorganisms. To increase the probability of detecting microorganisms, many laboratories performed pre-processing of the samples. The additional value of pre-processing tissues is still the matter of debate (Cai et al., 2021;Yusuf et al., 2021).
Synovial fluid culture is mentioned as a criterion by MSIS, IDSA, and EBJIS, but only IDSA requested that synovial fluid should be performed in all suspected PJI cases. In our survey, only half of the participants received synovial fluid. Our survey question was not specified to ask about synovial fluid prior to the surgery. Arguably, the number of synovial fluid prior to surgery may be even lower. Synovial fluid in a period before surgery is the only microbiological sample possible at that time, but it is laborious to perform, is associated with contamination risk and possible complication, and has limited predictive value (Schulz et al., 2021). Moreover, the volume varies and sometimes it is insufficient for culture. These factors may cause reluctancy to perform synovial fluid culture. When no synovial fluid culture is submitted, one of the possible criteria diagnose PJI is missing. Consequently, PJI rate will be lower than when synovial fluid culture is available for culture.
Approximately half of the participants in this survey received explanted prosthesis on a regular basis. While IDSA and EBJIS mentioned culture of explanted prosthesis as criteria, MSIS does not. There is also subtle difference between the positivity criteria of explanted foreign body culture between these organizations. While IDSA uses a qualitative criterion, EBJIS uses quantitative. In practice, quantitative criteria are prone to variations. In the original paper on sonication in PJI, 50 colony-forming units (CFU) per plate were used as cut-off, as

Culture requirements to be submitted
Intraoperative of tissues culture Yes: at least 3 and no more than 5 periprosthetic specimen culture samples be taken and incubated in an aerobic and anaerobic environment Yes: at least 3 and optimally 5 or 6 periprosthetic intraoperative tissue samples or the explanted prosthesis, aerobic and anaerobic culture Not specifically mentioned Preoperative synovial fluid Not specifically mentioned but is mentioned in diagnosis criteria.
Yes: should be performed in all patients with suspected acute PJI unless the diagnosis is evident clinically and surgery is planned and antimicrobials can be safely withheld prior to surgery.
Not specifically mentioned but is mentioned in diagnosis criteria.
Requirement for explanted prosthesis to be submitted for culture  (Trampuz et al., 2007). The protocol in this paper mentioned 400 ml Ringer's solution to be added to the container contained removed prosthetic joint, and vortexing was performed prior to and after sonication. The incubation time of the media was up to 7 days. Our survey shows that most of the laboratories do not follow this protocol such the use of pre-determined volume to be added in the container. Quantitative reporting will also depend on whether centrifugation is performed. Swabs are an inferior type of samples to make PJI diagnosis and should be discouraged. Fortunately, only 20-30% of the participants received swab samples in the majority of the cases. Swabs were more frequently used in settings with a larger distance between the hospital and the laboratory. Perhaps, clinicians consider swabs as a convenient sample when it needs to be transported for longer distances. An additional explanation could be a greater difficulty of communicational exchange.
The culture media were mostly incubated for 7-14 days in this survey. The prolonged incubation was recommended to detect slow growing bacteria, such as Cutibacterium acnes (Schäfer et al., 2008;Kheir et al., 2018). This duration is relevant for late PJI, but less for acute PJI. Yet, it is logistically impossible for a clinical laboratory to have separate protocol for acute and late PJI samples. A space constraint of blood culture incubation system due to prolonged incubation is also an issue. A possible intermediate approach is to prolong the incubation only if low grade infection possibility is communicated by the clinician to the laboratory. The authors are of the opinion not to extend the culture duration beyond 14 days due to possible contaminants.
Multidisciplinary meetings take places in the majority of the participant's hospitals. Often, these meetings take place in larger hospitals perhaps due to logistic reason. Such meetings should also be encouraged in the smaller hospitals. Clearly, success of PJI diagnosis and treatment needs close collaboration between orthopedic surgeons, clinical microbiology specialists, and infectious disease specialists (Yusuf and Borens, 2015).
The present study is the largest in-depth survey regarding PJI diagnostic, but there are several limitations. First, respondents Frontiers in Microbiology | www.frontiersin.org were selected by among colleagues who worked regularly with PJI diagnostics, and this may somehow introduce bias. However, this selection was needed due to the 'in depth' nature of this survey. The participants are well balanced between academic and public hospital clinical microbiologists. The geography of participants seems also representative (Figure 1). Secondly, this survey did not assess new and innovative tests, such as microcalorimetry and biomarkers, in synovial fluid. MSIS uses biomarker criteria, such as alpha-defensin. Thirdly, the survey questions were not designed to differentiate samples from various surgical procedures (debridement, antibiotic and implant retention or single or multi-stage joint replacement), or to differentiate acute from late PJI. Yet, it is practically impossible for a clinical microbiology laboratory to make such a differentiation. It is also tempting to perform analysis on the differences between countries since health care system may influence the results. However,various size and number of participants per countries did not allow comparison to be made.
The findings from this survey can have several consequences of our findings. First, clinical microbiologists should be involved in the diagnostic process of PJI because reported culture results are results of complex processes which need some interpretation. Second, this study gives indication on 'common' diagnostic procedure for PJI in Europe. These common practices may help for harmonization of diagnostics approaches, in order to improve comparability and interpretation of reported results. Third, any variation in the PJI studies regarding prevalence and definition of PJI shouldbe interpreted in light of various laboratory methods. The PJI definition criteria used may be the same, but the laboratory procedures may be different.
In conclusion, this survey shows similarity and differences of clinical microbiology tests and consultation for PJI diagnostics and therapy in several European countries.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

AUTHOR CONTRIBUTIONS
EY: conceptualization, methodology, formal analysis, investigation, resources, data curation, and writing-original draft. CR: conceptualization, methodology, resources, data curation, and writing-review and editing. JE, AR, AT, and PW: conceptualization, investigation, and writing-review and editing. TK: conceptualization, methodology, formal analysis, investigation, resources, data curation, and writing-review and editing. All authors contributed to the article and approved the submitted version.