O44 Universal genotyping of Mycobacterium tuberculosis in Ontario, Canada: observations from the OUT-TB Initiative

Abstract

Methods: The Platelia GM EIA assay (Bio-Rad) was performed on all BAL specimens obtained from haematology-oncology patients according to the manufacturer’s recommendations for serum. Cases were classified as proven, probable, possible or definitely no invasive pulmonary aspergillosis (IPA) according to the Revised Definitions of Invasive Fungal Disease from the EORTC/MSG Consensus Group. For performance characteristics, proven and probable cases were considered as IPA; possible cases were considered as without IPA. The result of BAL GM was not considered as a criterion to classify cases in order to avoid incorporation bias. In patients with more than one positive (GM index 0.5) specimen, only the first was considered for the analysis. Data were analyzed and 95% confidence intervals (CI) calculated with Stata 8.0. Results: Between March 2005 and April 2008, 173 BAL samples from 145 patients were included in this study. We identified 5 proven, 7 probable (total of 12 IPA cases; 6.9%) and 35 possible cases of IPA. Using an index cut-off of 0.5, BAL GM assay was positive in 47 (27.2%) specimens. Sensitivity and specificity of the GM assay in BAL were 100% (95%CI 73.5 100) and 78.3% (95%CI 71.1 84.4), respectively; positive predictive value (PPV) and negative predictive value (NPV) were 25.5% (95%CI 13.9 40.3) and 100%, respectively. False-positive results were found in 21 patients with definitely no IPA (index ranged from 0.5 to 6.2) and in 14 with possible IPA (index ranged from 0.6 to 7.1). All 12 cases of probable or proven IPA had GM index cut-off of 3.0 (range 3.0 8.8). Using a GM index cut-off of 2.0, 24 (13.9%) specimens were positive: 12 from patients with probable or proven IPA, 5 from patients with possible IPA and 7 with definitely no IPA. Sensitivity remained 100% but specificity increased to 92.5% (95%CI 87.3 96.1); PPV increased to 50.0% (95%CI 29.1 70.1) and NPV remained at 100%. Conclusions: The clinical utility of GM assay in BAL mainly lies in its NPV, thus identifying patients at low risk of IPA. Increasing cut-off value from 0.5 to 1.0 has been suggested by other investigators. We argue for increasing BAL GM index cut-off value even higher to 2.0; this would increase significantly the specificity, without decrease in sensitivity.