BackgroundInfections are the major cause of morbidity and mortality in children with cancer. Gaining a favorable prognosis for these patients depends on selecting the appropriate therapy, which in turn depends on rapid and accurate microbiological diagnosis. This study employed real-time PCR (qPCR) to identify the main pathogens causing bloodstream infection (BSI) in patients treated at the Pediatric Oncology Institute IOP-GRAACC-UNIFESP-Brazil. Antimicrobial resistance genes were also investigated using this methodology.MethodsA total of 248 samples from BACTEC® blood culture bottles and 99 whole-blood samples collected in tubes containing EDTA K2 Gel were isolated from 137 patients. All samples were screened by specific Gram probes for multiplex qPCR. Seventeen sequences were evaluated using gender-specific TaqMan probes and the resistance genes blaSHV, blaTEM, blaCTX, blaKPC, blaIMP, blaSPM, blaVIM, vanA, vanB and mecA were detected using the SYBR Green method.ResultsPositive qPCR results were obtained in 112 of the blood culture bottles (112/124), and 90 % agreement was observed between phenotypic and molecular microbial detection methods. For bacterial and fungal identification, the performance test showed: sensitivity 87 %; specificity 91 %; NPV 90 %; PPV 89 % and accuracy of 89 % when compared with the phenotypic method. The mecA gene was detected in 37 samples, extended-spectrum β-lactamases were detected in six samples and metallo-β-lactamase coding genes in four samples, with 60 % concordance between the two methods. The qPCR on whole blood detected eight samples possessing the mecA gene and one sample harboring the vanB gene. The blaKPC, blaVIM, blaIMP and blaSHV genes were not detected in this study.ConclusionReal-time PCR is a useful tool in the early identification of pathogens and antimicrobial resistance genes from bloodstream infections of pediatric oncologic patients.

Tuberculous meningitis (TBM) is a severe form of extrapulmonary tuberculosis. The aims of this study were to evaluate in-house molecular diagnostic protocols of DNA extraction directly from CSF samples and the targets amplified by qPCR as an accurate and fast diagnosis of TBM. One hundred CSF samples from 68 patients suspected of TBM were studied. Four DNA extraction techniques (phenol-chloroform-thiocyanate guanidine, silica thiocyanate guanidine, resin, and resin with ethanol) were compared and CSF samples were used to determine the best target (IS6110, MPB64, and hsp65 KDa) by qPCR. The extraction protocol using the phenol-chloroform-thiocyanate guanidine showed the best results in terms of quantification and sensitivity of PCR amplification, presenting up to 10 times more DNA than the second best protocol, the silica guanidine thiocyanate. The target that showed the best result for TBM diagnosis was the IS6110. This target showed 91% sensitivity and 97% specificity when we analyzed the results by sample and showed 100% sensitivity and 98% specificity when we analyzed the results by patient. The DNA extraction with phenol-chloroform-thiocyanate guanidine followed by IS6110 target amplification has been shown to be suitable for diagnosis of TBM in our clinical setting.