How predictable is development of resistance after beta-lactam therapy in Enterobacter cloacae infection?

Certain non-fastidious Gram-negative bacilli, notably Enterobacter cloacae, although classified as susceptible by usual in-vitro susceptibility testing, often become resistant in patients treated with newer beta-lactam antibiotics. Here various in-vitro tests were carried out together with an animal model allowing the quantification of resistance that emerges after short term therapy. Mice were challenged (10(8) cfu plus talcum) intraperitoneally with one each of four strains of Ent. cloacae. Two hours later, a single beta-lactam dose was administered subcutaneously. The following day, the peritoneal bacterial population was analysed by using antibiotic-containing gradient plates. Development of resistance after therapy varied according to the compound considered. Imipenem (50 mg/kg) produced no resistance, and piperacillin (200 mg/kg) only a few, while resistance occurred frequently after therapy with aztreonam (50 mg/kg), ceftazidime (50 mg/kg), cefotaxime (50 mg/kg) and cefpirome (50 mg/kg). MICs increased by at least 16-fold when resistance developed. No simple correlations were found between these in-vivo results and initial MICs, killing kinetics, frequency of resistant variants within the bacterial populations before therapy, initial MIC of these variants or antibiotic concentrations assayed in peritoneal fluid 60 min after dosing. The most reliable predictive in-vitro test appeared to be the determination of resistance emerging in broth containing at least 16 times the MIC of the antibiotic tested. Such a test is unlikely to be used on a routine basis. When a beta-lactam compound seems appropriate for treating an Enterobacter infection, it may be advisable to avoid drugs that are prone to produce resistance in experimental or clinical infections, whatever the results of conventional in-vitro susceptibility tests.