Thermogenic diagnosis of periprosthetic joint infection by microcalorimetry of synovial fluid

Our institutional review board issued its approval for this prospective study (EA1/306/14) and written informed consent was obtained for all patients. This study is compliant with the Declaration of Helsinki and was registered under the public clinical trial identification NCT02530229 (https://​clinicaltrials.​gov/​ct2/​show/​NCT02530229).

Patients aged 18 years and older presenting with a painful prosthesis underwent joint aspiration between December 2014 and November 2015 in a tertiary healthcare facility as part of routine preoperative diagnosis. Patients with an aspirated synovial fluid volume below 5 ml, joint aspiration within 6 weeks of the last surgery of the affected joint, periprosthetic fracture and prothesis dislocation were excluded from this study.

At least one of the following criteria were necessary for diagnosing PJI, as used in other studies [5, 9, 10, 17‐19]: (i) visual identification of macroscopic purulence around the prosthesis, (ii) a visible sinus tract, (iii) an elevated synovial fluid leukocyte count (> 2000 leukocytes/μl) or differential (> 70% granulocytes) [3, 4], (iv) positive culture of synovial fluid, periprosthetic tissue or culture of sonication, (v) histopathologic inflammation with ≥2 granulocytes per high-power field in periprosthetic tissue (corresponding type II or III according to Krenn and Morawietz [20]).

Positivity of synovial fluid culture was considered for a confined microbial organism in solid media culture. For enrichment broth, positivity was contemplated only for high-virulent pathogens. Positivity for culture of periprosthetic tissue was defined for (i) growth of one or more highly virulent pathogens (Enterobacterspp., Streptococcus, Candida, Staphylococcus aureus) or (ii) growth of at least two low virulent organisms (coagulase-negative staphylococci, Cutibacterium spp., enterococci, and pathogens of the regular skin microbioma) [21]. Growth of a high-virulent pathogen in sonication fluid of one or more colony forming unit (CFU)/ml or growth of > 50 CFU/ml of a low-virulent pathogen were required to confirm positivity for sonication [22].

Asepsis was applied to the skin before joint aspiration. The corresponding joint was aspirated with a sterile 18-gauge spinal needle in consonance with aseptic directives. Synovial fluid was aspirated into a 10-ml syringe. The needle was relocated inside the joint without trespassing the skin if the initial joint aspiration was void.

A pediatric blood culture bottle (BacTec PedsPlus/F, Beckton Dickinson&Co., IE) with at least 1 ml of synovial fluid was employed for culture (at 37 °C) and monitored for 14 days (or until positivity). For gram stain an additional minimum of 1 ml was applied. Further explorations comprised Schaedler, chocolate and blood agar plates, and thioglycolate broth. Aerobic and anaerobic cultures at 37 °C were monitored daily for 7 and 14 days, respectively. Exploration for pyrophosphate and urate crystals and inflammatory cells was performed with polarization microscopy on an additional 1 ml of synovial fluid. An ethylenediaminetetraacetic acid (EDTA) vial was filled with at least 1 ml of synovial fluid to determine leukocyte count and differential. 10 μl of hyaluronidase (Sigma-Aldrich, DE) was applied on clotted samples at room temperature conditions for a 10 min-period.

For microcalorimetric analysis, an aliquot of 1 ml synovial fluid was inoculated into a 4 ml microcalorimetry vial pre-filled with 2 ml of tryptic soy broth. The vial was inserted into an isothermal microcalorimeter with 48 measuring channels (model 3102 TAM III, TA Instruments, New Castle, DE, USA). According to the manufacturer, the sensitivity of the microcalorimeter is 0.225 μW. First, an equilibration process was performed for 15 min in order to reach 37 °C and another 45 min to get accurate measurement of heat flow. Then, heat production was continuously measured for a minimum of 5 days, expressed in μWatts over time. Calorimetry heat peak values were measured as the difference between the maximum heat value of the curve and its baseline value (Fig. 1). Baseline was defined as the heat value approximated by the asymptote at the end of the timeline. The time to positivity for microcalorimetry was determined and was defined as the time elapsed between insertion of the ampoule into the microcalorimeter and a rising heat curve reaching a minimum threshold of 10 μW compared to baseline [14, 15]. Upon termination of microcalorimetry testing, aerobic and anaerobic agar plating was performed with the remaining microcalorimetric probe.

Cultures and histopathological examination were performed on periprosthetic tissue obtained in revision surgery. For intra-operative diagnostics periprosthetic tissue samples were collected by the surgeon and sent in for culture (at least 4 samples) and histopathologic examination (at least 1 sample). Similar to the aforementioned processing of synovial fluid, homogenization and culture of periprosthetic tissue samples was performed on aerobic and anaerobic agar plates and thioglycolate broth. Microbial pathogens were automatically determined according to standard procedure (VITEK 2, BioMérieux, FR). Sonication of the extracted implant was performed following standard technique [18].

The diagnostic methods were evaluated with McNemar’s Chi-squared test (values at p < 0.05 were considered statistically significant). For aseptic calorimetry curves, we explored the association between leukocyte count in synovial fluid and heat production with Pearson’s correlation coefficient r (strong correlation for r > 0.8, modest for r = 0.5–0.8, and weak for r < 0.5). The relationship between leucocyte count and peak heat production was explored with linear regression.

Of 135 evaluated patients, 107 patients were included in this study. Excluded were 11 patients with insufficient synovial fluid (of whom 10 had a hip and 1 knee prosthesis), 5 with aspiration within 6 weeks of surgery, 4 patients with periprosthetic fracture and 8 patients with a prosthesis dislocation. The demographic characteristics of included patients is shown in Table 1. A total of 69 patients (65%) underwent revision surgery. For these patients, additional intra-operative samples were obtained for microbiological and histopathological analysis, as well as sonication of the extracted prosthesis.

Of 46 (43%) patients diagnosed with PJI, the infecting microorganism was identified in 26 (56%) cases. The pathogen was isolated in 18 (39%) PJI cases by synovial fluid culture only. The pathogen was detected in an additional 8 cases with the addition of periprosthetic tissue and sonication fluid cultures. Of these 8 patients with negative synovial fluid cultures, 6 cases had positive periprosthetic tissue cultures, 4 cases had positive sonication cultures and 2 cases had both. Hence, 26 (56%) PJI cases had any positive culture.

For 20 PJI cases (44%), no pathogen was identified as all (synovial fluid, periprosthetic tissue and sonication fluid) cultures remained negative.

Out of the 18 identified pathogens in synovial fluid culture, the most recurrent causative organisms detected were Streptococcus spp. in 8 (44%) cases, coagulase-negative staphylococci in 3 (17%) cases, Escherichia coli in 3 (17%) and Enterococcus spp. in 3 (17%) cases. No polymicrobial infections were detected by synovial fluid culture.

The results for the diagnostic methods are listed in Table 2 and positive calorimetry curves are shown in Fig. 2. Synovial fluid leukocytes (absolute leukocyte count or percentage of granulocytes) was the most sensitive diagnostic method for the diagnosis of PJI (80%), followed by periprosthetic tissue histopathology (74%). The sensitivity of synovial fluid microcalorimetry and culture was both 39%. The overall accuracy of all pre- and post-operative cultures (synovial fluid, periprosthetic tissue and sonication fluid) was 81%. The median time until positivity for microcalorimetry was 9 h (range, 1–64 h).

In 98 of 107 patients (92%), synovial fluid culture and microcalorimetry showed concordant results, including 38 PJI and 60 aseptic cases (Table 3). In patients with PJI, microcalorimetry missed 4 cases where cultures grew the pathogen, including E. faecalis (2 isolates), coagulase-negative staphylococci and E. coli (1 isolate each). In contrast, microcalorimetry was positive for 5 cases with negative synovial fluid culture (Table 4). Culture of synovial fluid after microcalorimetric measurement was performed in 68 cases (64%) (Table 4). Microcalorimetry was positive in one aseptic failure case with negative synovial fluid culture. The agar plating of this microcalorimetric probe was negative and the case was considered as false-positive.

Pearson’s correlation coefficient for log leucocyte count and log peak heat production showed a weak linear correlation coefficient of r = 0.366 with a 95% confidence interval 0.130–0.567 and a significance of p < 0.001 (Fig. 3). Linear regression showed that a 1% increase in leukocyte count is associated with a 0.72% increase in peak heat production (p = 0.003).