Performance of the galactomannan antigen detection test in the diagnosis of invasive aspergillosis in children with cancer or undergoing haemopoietic stem cell transplantation.

Serum galactomannan (GM) antigen detection is not recommended for defining invasive aspergillosis (IA) in children undergoing aggressive chemotherapy or allogeneic haemopoietic stem cell transplantation (HSCT). The ability of the GM test to identify IA in children was retrospectively evaluated in a cohort of children. Test performance was evaluated on samples that were collected during 195 periods at risk of IA. Proven IA was diagnosed in seven periods, all with positive GM test results (true positives, 4%), and possible IA was diagnosed in 15 periods, all with negative GM test results (false negatives, 8%). The test result was positive with negative microbiological, histological and clinical features in three periods (false positives, 1%), and in 170 periods it was negative with negative microbiological, histological and clinical features (true negatives, 87%). The sensitivity was 0.32 and the specificity was 0.98; the positive predictive value was 0.70 and the negative predictive value was 0.92. The efficiency of the test was 0.91, the positive likelihood ratio was 18.3, and the negative likelihood ratio was 1.4. The probability of missing an IA because of a negative test result was 0.03. Test performance proved to be better during at-risk periods following chemotherapy than in periods following allogeneic HSCT. The GM assay is useful for identifying periods of IA in children undergoing aggressive chemotherapy or allogeneic HSCT.