Authors: Elie Ghanem, MD; James J Purtill, MD; Peter F Sharkey, MD; Robert Barrack, MD; * Javad Parvizi, MD*
Title: Diagnostic Dilemma for Periprosthetic Infection
Addresses: Rothman Institute, 925 Chestnut Street, Philadelphia, Pa 19107
Purpose: Periprosthetic infection (PPI) remains the most dreaded and difficult complication of total knee arthroplasty (TKA). Although there is no absolute diagnostic test for PPI, synovial leukocyte counts and neutrophil percentages have been reported to have high sensitivity and specificity making them valuable tests for diagnosis. However, no cut off value for the latter is agreed upon. This study intends to define the cut-off values for preoperative synovial fluid leukocyte count and PMN percentage that allow physicians to diagnose infection in a prosthetic joint.
Methods: We retrospectively analyzed synovial fluid aspirated preoperatively from 593 total knee arthroplasties (161 infected; 432 aseptic) from three different institutions. Using ROC curves, we determined cut-off values with optimal accuracy in diagnosis of infection for fluid leukocyte count and PMN percentage. The diagnosis of infection was based on the fulfillment of 3 out of 5 criteria: positive preoperative aspiration culture, positive intraoperative culture, purulent intraoperative tissue appearance, ESR >30mm/hr and CRP >1mg/dl. The sensitivity, specificity, and predictive values were calculated for the above cut-off values. Chi-square analysis was used to determine if an association existed between PPI and synovial leukocyte count and neutrophil percentage.
Results: The synovial fluid leukocyte count was higher in patients with PPI (median, 31 x 103 cell/µl; range, 0.308 x 103 cells/µl to 695 x 103 cells/µl) compared to aseptic joint arthroplasties (median, 0.219 x 103 cells/µl; range, 0.040 x 103 cells/µl to 1.76 x 103 cells/µl; p<0.0001). Similarly, the neutrophil percentage was higher in patients with prosthetic joint infection (median, 90%; range, 65% to 98%) than in those with noninfected joints (median, 26%: range, 2% to 83%; p<0.0001). The cut-off values for optimal accuracy in diagnosis of infection were 1760 cells/µl for fluid leukocyte count and 73% for PMN percentage. A fluid cell count >1760 cells/µl had a PPV of 99% and NPV of 88%, while a neutrophil percentage >73% yielded a PPV of 96% and NPV of 91%. Chi-square analysis revealed the fluid WBC count and the neutrophil percentage to be significantly associated with the diagnosis of infection (p<0.0001).
Discussion: A recent study by Trampuz et al. observed a PPV of 73% and NPV of 98% when a cut-off value of 1700 cells/µl for fluid leukocyte was used, while >65% fluid neutrophils yielded a PPV of 94% and NPV of 99%. Our study indicated that a fluid WBC cut-off value of 1760 cells/µl yielded better positive (99%) predictive value but slightly less negative predictive values (88%). However, the cut-off value of our PMN percentage was slightly higher (73%) and had similar PPV (96%), but slightly less NPV (91%).
Synovial fluid leukocyte count and differential are useful adjuncts to ESR and CRP in the preoperative diagnosis for infection in total knee arthroplasties. Both tests can reliably confirm infection if the screening tests are positive.