Background: Bloodstream infections (BSIs) are the major cause of mortality in cancer patients. Molecular techniques are used for rapid diagnosis of BSI, allowing early therapy and improving survival. We aimed to establish whether real-time quantitative polymerase chain reaction (qPCR) could improve early diagnosis and therapy in paediatric cancer patients, and describe the predominant pathogens of BSI and their antimicrobial susceptibility.
Methods: Blood samples were processed by the BACTEC system and microbial identification and susceptibility tests were performed by the Phoenix system. All samples were screened by multiplex 16 s rDNA qPCR. Seventeen species were evaluated using sex-specific TaqMan probes and resistance genes blaSHV, blaTEM, blaCTX, blaKPC, blaIMP, blaSPM, blaVIM, vanA, vanB and mecA were screened by SYBR Green reactions. Therapeutic efficacy was evaluated at the time of positive blood culture and at final phenotypic identification and antimicrobial susceptibility results.
Results: We analyzed 69 episodes of BSI from 64 patients. Gram-positive bacteria were identified in 61 % of the samples, Gram-negative bacteria in 32 % and fungi in 7 %. There was 78.2 % of agreement between the phenotypic and molecular methods in final species identification. The mecA gene was detected in 81.4 % of Staphylococcus spp., and 91.6 % were concordant with the phenotypic method. Detection of vanA gene was 100 % concordant. The concordance for Gram-negative susceptibilities was 71.4 % for Enterobacteriaceae and 50 % for Pseudomonas aeruginosa. Therapy was more frequently inadequate in patients who died, and the molecular test was concordant with the phenotypic susceptibility test in 50 %.
Conclusions: qPCR has potential indication for early identification of pathogens and antimicrobial resistance genes from BSI in paediatric cancer patients and may improve antimicrobial therapy.