Aim: To compare PCR with galactomannan antigen detection for the diagnosis of invasive aspergillosis (IA).
Methods: We prospectively collected serial blood samples from haematological patients at risk of IA, and analysed their samples retrospectively for galactomannan (GM) antigen using the Platelia test and for aspergillus DNA using an in-house PCR-ELISA assay. Matched GM and PCR analyses were performed on 263 samples from 25 patients. Patients were classified for potential IA according to international consensus criteria, with five patients classified as positive (four proven, one probable) and 20 classified as negative (seven possible, 13 no evidence IA).
Results: All five patients with IA were positive by PCR with positive results in 24 of 82 samples, whereas three of five patients were positive by GM with four of 82 samples being positive. Three of 20 patients without IA were positive by PCR in 18 of 181 samples, whereas corresponding results for GM detection were one of 20 and one of 181, respectively. Adjustment of ELISA cut-off values and/or the requirement for two consecutive samples to be positive generated different results; however, lowering the positivity index (PI) for GM detection to 0.5 did not improve the sensitivity of the assay. Optimal results for PCR detection and GM were: 100% and 60% sensitivity, 85% and 95% specificity, 0.625 and 0.75 positive predictive value, and 1.0 and 0.8 negative predictive value, with a false-positive sample rate of 8 and 0.4%, positive likelihood ratio of 6.66 and 11.99 and negative likelihood ratio of 0 and 0.42, respectively.
Conclusions: This PCR method is very sensitive for the diagnosis of IA but is associated with a moderate rate of false positives; the GM assay exhibited poor sensitivity but high specificity. Further evaluation of PCR assays for the diagnosis of IA and other invasive fungal infections is warranted.