Methods: The study was approved by the Institutional Review Board. All patients undergoing treatment for leukemia at St. Jude Children’s Research Hospital with CoNS isolated from blood between 2010 and 2016 were eligible. The first available isolate from each blood culture episode was obtained from the clinical laboratory and tested for vancomycin heteroresistance by population analysis profiling in comparison to the hVISA strain Mu3. Clinical data were collected from the medical record for up to 9 months after the episode. Episodes with ≥2 positive cultures or a single positive culture from a single lumen CVC were classified as CLABSI. Outcomes of interest included Treatment Failure (death or relapse of infection) or Poor Response to vancomycin therapy (persistence of bacteremia ≥1 day after initiation of vancomycin or Treatment Failure). Logistic regression was used to test associations between heteroresistance and exposures, and cumulative incidence analyses were used to test the effect on outcomes.
A total of 74 CoNS isolates were obtained from 65 participants, 39 with ALL and 26 with AML; 25/74 (33.8%) of isolates showed showed heteroresistance. The strongest identified risk factor for infection with a heteroresistant organism was number of days of vancomycin in the preceding 60 days (OR=1.05/day; P=0.035). In the 40 CLABSI episodes, heteroresistant isolates had a higher cumulative incidence of Poor Response and of Treatment Failure (P=0.006 and P=0.003 respectively).
Vancomycin heteroresistance is common in CoNS causing CLABSI in children undergoing treatment for leukemia, and is associated with an increased risk of Treatment Failure. Further research should aim to validate this finding in an independent cohort and identify strategies to improve the diagnosis and treatment of these infections.
R. Hayden, Roche Molecular: Scientific Advisor , Consulting fee . Abbott Molecular: Scientific Advisor , Consulting fee . Quidel: Scientific Advisor , Consulting fee .
J. Rubnitz, None
J. Wolf, Karius Inc.: Investigator , Research support .
J. Rosch, None