Can we rely on the combination of serological tests and frozen sections at the time of reimplantation for two-stage exchange hip arthroplasty in patients with a “dry tap”?

After Institutional Review Board approval, we retrospectively reviewed 129 patients (129 hips) who met the Musculoskeletal Infection Society (MSIS) criteria for PJI [15] and underwent a two-stage exchange arthroplasty between 2012 and 2016. Patients with megaprosthesis, prior two-stage exchange arthroplasty, spacer exchange in the interim, and patients with missing critical data were excluded. Then 97 two-stage exchanges were included in the final analysis.

Serum ESR and CRP values were routinely obtained before second-stage reimplantation in our institution. ESR and CRP levels were measured by a BN™ II System (Siemens, Marburg, Germany) and Automatic Sed-rate Analyzer 20/100 (VACUETTE) SRS/100 (Greiner Bio-One GmbH, Kremsmunster, Austria), respectively. Other clinical records of these patients were manually reviewed, including age, gender, body mass index (BMI), intraoperative cultures, and histological analyses at the time of reimplantation and follow-up data.

An institutional standard protocol of two-stage exchange arthroplasty was performed. During the first-stage resection arthroplasty, all implanted components were removed followed by extensive debridement and irrigation. An antibiotic-loaded cement spacer, containing 6–10 g of vancomycin and 2–4 g of meropenem, was then inserted. The combination of vancomycin and meropenem in the bone cement was utilized in accordance with our institutional infection control department, which explained that more than 90% of the organisms isolated from patients with PJI in our institution were sensitive to one or both antibiotics. Following resection arthroplasty, at least 6 weeks of systemic antibiotic therapy were prescribed. The selection of antibiotic application was based on culture sensitivity reports and institutional guidelines infectious specialists’ consultation. Reimplantation was performed after 2–4 weeks of antibiotic holiday, and the clinical course presented no signs of infection. During the second stage, the antibiotic-loaded cement spacer was removed and new prostheses were reimplanted followed by re-debridement and irrigation. Prophylactic antibiotics were continued for another 5 days after reimplantation.

Intraoperative frozen sections were routinely taken at the time of reimplantation. Three to five samples of tissues for frozen sections were obtained during surgery from the periprosthetic membrane and other periprosthetic tissues in which infection was suspected. Each sample was gathered in a separate clean specimen bag and was promptly referred to the pathology department in a sterile transport to avoid cross-contamination. All samples were stained using hematoxylin and eosin (H&E) and analyzed based on Feldman et al.’s criteria [16]. Multiple sections from each sample were classified by two experienced pathologists, and the number of PMNs per HPF (×400) was determined in 5–10 separate microscopic fields. The average was calculated as the result of the frozen section.

As hip aspirations were not routinely obtained prior to the second-stage reimplantation in our institution, the diagnosis of persistent infection was based on a modified MSIS criteria [8, 17]: the presence of a sinus tract communicating with the joint at surgery or two positive intraoperative periprosthetic cultures with the same organism or fulfill two of three following minor criteria including (1) an elevated erythrocyte sedimentation rate [ESR > 30 mm/h] and C-reactive protein [CRP > 10 mg/L], (2) a single positive intraoperative periprosthetic tissue culture, and (3) a positive histologic analysis of periprosthetic tissue [> 5 neutrophils per high power field]. Additionally, persistent infection before reimplantation was also diagnosed by treatment failure following reimplantation in the sensitivity analysis (details below).

All of the statistical analyses were performed with the statistical software packages R (http://​www.​R-project.​org, The R Foundation). Categorical variables were presented as frequencies and percentages, and continuous variables as means and standard deviation. The differences between infection and non-infection groups were compared with the use of the Mann-Whitney test for continuous variables and the Fisher’s exact test for categorical variables. Two diagnostic models for assessment of persistent infection before reimplantation were developed, with model 1 using serum ESR and CRP with the combination of frozen section > 5 PMNs as a cutoff value and model 2 using ESR and CRP with the combination of frozen section > 10 PMNs as a cutoff value. Receiver operating characteristic (ROC) curves were generated using bootstrap resampling (times = 500) to determine the predictive values of the two models. The area under the ROC curve (AUC) with 95% confidential interval (CI) was used as a measure of diagnostic accuracy. Then the two models were compared using the DeLong method [18]. The area under the ROC curve (AUC) with 95% CI was calculated. Discriminatory value of ROC curves was interpreted as excellent (AUC 0.9–1), good (0.8–0.89), fair (0.7–0.79), poor (0.6–0.69), or fail/no discriminatory capacity (0.5–0.59) [19]. A p value of 0.05 was considered significant.

Given no “gold standard” in predicting persistent infection before reimplantation, a set of sensitivity analysis was conducted by using treatment success following reimplantation as a proxy for infection eradication at the time of reimplantation. Treatment success was defined according to the Delphi consensus criteria proposed by Diaz-Ledezma et al. [20]: (1) infection eradication characterized by a healed wound without drainage, fistula, or pain, with no recurrence of infection; (2) no occurrence of periprosthetic joint infection-related mortality (e.g., sepsis, necrotizing fasciitis); or (3) no subsequent surgical intervention for infection after reimplantation surgery.

Only 86 patients with a minimum follow-up of 1 year were included in the sensitivity analysis. Kaplan-Meier (KM) survivorship curves were generated for follow-up at 1, 2, and 5 years. Two diagnostic models for assessment of persistent infection before reimplantation were developed, with model 3 using serum ESR and CRP with the combination of frozen section > 5 PMNs as a cutoff value and model 4 using ESR and CRP with the combination of frozen section > 10 PMNs as a cutoff value. The ROC curves of model 3 and model 4 were generated and compared according to the methods above.

According to the Delphi criteria, the overall treatment success rate was 87.2% (75/86). There was no PJI-related mortality. The survivorship with treatment success as an endpoint was 93.0% (95% CI, 87.8 to 98.6%) at the 1-year follow-up, 91.9% (95% CI, 86.3 to 97.8%) at the 2-year follow-up, and 89.1% (95% CI, 81.6 to 97.2%) at the 5-year follow-up (Fig. 2).

In the sensitivity analysis, eleven patients with treatment failure were considered as persistent infection at the time of reimplantation. The formulas for model 3 and model 4 were shown in Table 3. The results of sensitivity analysis were consistent with above results as the AUC for model 3 and model 4 was 0.669 (95% CI, 0.532–0.807) and 0.668 (95% CI, 0.486–0.832), respectively (Fig. 3). The performances were depicted in Table 4, and there was no significant difference between models 3 and 4 (p = 0.989).