Performance of penicillinase detection tests in Staphylococcus epidermidis: comparison of different phenotypic methods.

Staphylococcus epidermidis is the leading coagulase negative staphylococci (CoNS) species associated with healthcare associated infections. In order to de-escalate antimicrobial therapy, isolates of S. epidermidis lacking the blaZ gene should be eligible for targeted antimicrobial therapy. However, testing the susceptibility of coagulase negative staphylococci (CoNS) to penicillin G is no longer recommended by EUCAST, given the low performances for penicillinase detection in CoNS. The objective of this work was to determine a phenotypic method with high performance for detecting penicillinase production in S. epidermidis. Four techniques for the detection of penicillinase production (disk diffusion, zone edge test, nitrocefin test, Minimal Inhibitory Concentration (MIC) by automated system Vitek2®) were evaluated on 182 S. epidermidis isolates, using identification of blaZ gene by PCR as the reference method. The performance of the methods for penicillinase detection was compared by the sensitivity, the specificity, the negative predictive value and the positive predictive value, and with Cohen’s kappa statistical test. Among the 182 S. epidermidis included in this study, 55 carried the blaZ gene. The nitrocefin test, characterized by a poor sensitivity (91%), was therefore excluded from S. epidermidis penicillinase detection. The algorithm proposed here for the penicillinase detection in S. epidermidis involved two common antimicrobial susceptibility techniques: disk diffusion method and MIC by Vitek2® system. Disk diffusion method, interpreted with a 26 mm breakpoint for penicillin G, was associated with a high sensitivity (98%) and specificity (100%). This method was completed with zone edge test for S. epidermidis with penicillin G diameter from 26 to 35 mm (sensitivity of 98%). The Vitek2® system is associated with a low sensitivity (93%) and a high specificity (99%) This low sensitivity is associated with false negative results, in isolates with 0.12 mg/L Penicillin G MIC values and blaZ positive. Thus for penicillin G MIC of 0.06 mg/L or 0.12 mg/L, a second step with disc diffusion method is suggested. According to our results, the strategy proposed here allows the interpretation of penicillin G susceptibility in S. epidermidis isolates, with an efficient detection of penicillin G resistance.