Development and validation of a clinical model to predict the presence of β-lactam resistance in viridans group streptococci causing bacteremia in neutropenic cancer patients

Background. Concern for serious infection due to β-lactam-resistant viridans group streptococci (VGS) is a major factor driving empiric use of an anti-gram-positive antimicrobial in patients with febrile neutropenia. We sought to develop and validate a prediction model for the presence of β-lactam resistance in VGS causing bloodstream infection (BSI) in neutropenic patients. Methods. Data from 569 unique cases of VGS BSI in neutropenic patients from 2000 to 2010 at the MD Anderson Cancer Center were used to develop the clinical prediction model. Validation was done using 163 cases from 2011 to 2013. In vitro activity of β-lactam agents was determined for 2011-2013 VGS bloodstream isolates. Results. In vitro resistance to β-lactam agents commonly used in the empiric treatment of febrile neutropenia was observed only for VGS isolates with a penicillin minimum inhibitory concentration (MIC) of ≥2 μg/mL. One hundred twenty-nine of 732 patients (17%) were infected with VGS strains with a penicillin MIC ≥2 μg/mL. For the derivation and validation cohorts, 98% of patients infected by VGS with a penicillin MIC of ≥2 μg/mL had at least 1 of the following risk factors: current use of a β-lactam as antimicrobial prophylaxis, receipt of a β-lactam antimicrobial in the previous 30 days, or nosocomial VGS BSI onset. Limiting empiric anti-gram-positive therapy to neutropenic patients having at least 1 of these 3 risk factors would have reduced such use by 42%. Conclusions. Simple clinical criteria can assist with targeting of anti-gram-positive therapy to febrile neutropenic patients at risk of serious β-lactam-resistant VGS infection.