The value of the serum aspergillus galactomannan (GM) to diagnose invasive aspergillosis (IA) and invasive fungal infections (IFI) as defined by European Organization of Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) in recipients of hematopoietic stem cell transplants (HSCT).

Background: Detection of serum GM is widely used and is incorporated into the definitions for diagnosis of IFI. There has been wide variability in its reported sensitivity based on the population tested and cut-off used.

Methods: We retrospectively analyzed data from all patients who had ≥1 GM after HSCT from December 2006 to September 2015. Cut-off of ≥0.5 optical density index was used to define a positive test. Chart review and review of institutional informatics database was performed to ascertain whether patients had possible, probable or proven IA and IFI per EORTC/MSG definitions at the time of the test. We excluded tests reported as invalid, tests for a patient after the first positive, duplicate negative tests done after diagnosis of at least a possible IFI, and positive tests in patients who had insufficient data to ascertain whether they had IFI. We collected data for receipt of semisynthetic penicillins in the week prior to a positive test.

Results: We identified 832 tests performed in 584 patients. Only 34/832 (4.1%) tests were positive, of which 27 were in patients with at least possible IA. 5/7 false positive tests were in patients infected with other molds known to cause a falsely positive test. No patient with a positive test had received any semisynthetic penicillin within a week prior to the test. 723/798 (90.6%) negative tests were in cases who did not satisfy criteria for at least a possible IFI at the time of the test. Sensitivity of the assay to detect at least possible IA was 26.5% (18.2 - 36.1%); specificity was 99.0% (98.3 - 99.6%); positive predictive value was 79.4% (62.1 - 91.3%) and negative predictive value was 90.6% (88.4 - 92.6%). When analysis was limited to cases of proven IFI, sensitivity to detect proven IA rose to 60% (26.2 - 87.9%). 1/162 tests in patients with an autologous HSCT and 33/637 tests in patients with an allogeneic HSCT were positive (p=0.008).

Conclusion: Serum GM was widely overused for diagnosis of IA at our institution. Limiting its use to patients with higher pre-test probability of an IFI increases its sensitivity. For HSCT recipients, we suggest performing the test only in allogeneic HSCT recipients who satisfy criteria for at least possible IFI per EORTC/MSG; this strategy would have missed only 1 positive test in our cohort.