Aspergillus Specific PCR and Galactomannan of Bronchoalveolar Lavage Are Superior to Concomitant Same-Day Testing of Concurrent Blood Samples in Immunocompromised Hematological Patients with Suspected Invasive Aspergillosis

Background
Invasive pulmonary aspergillosis (IPA) is a frequent complication in hematologic patients (pts) and proper diagnosis is difficult as culture based methods are time consuming and show only modest sensitivity, especially under the current practice of early mold-active antifungal therapy (AFT) or prophylaxis. Analyzing specimens representing the specific site of infection, i.e. bronchoalveolar lavage (BAL), is supposed to be superior compared to testing blood samples. However, that has not yet been shown in direct comparison in individual patients within a 24h time window.
To further elucidate this issue the diagnostic performance of an established Aspergillus specific nested Polymerase Chain Reaction (PCR) assay and of Galactomannan antigen testing (GM ELISA) in BAL was evaluated and compared to results obtained from concurrent blood samples of immunocompromised patients, most of all with hematological malignancies.

Methods:
We retrospectively identified 125 immunocompromised pts suspected to harbor IPA in whom BAL (n=92) and concurrent serum samples (within 24 hours) had been analyzed with Aspergillus PCR. Of these, 30/125 pts (24%) had proven/probable IPA according to 2008 EORTC/MSG criteria while the remaining 95 were classified as possible (n=65) or no IPA (n=30). Furthermore we also identified pts with same-day BAL and serum samples that had been analyzed with GM ELISA. Among those were another 26 pts that could be classified as proven/ probable IPA.

Results
PCR positivity in samples directly from the site of infection (BAL) was observed for 19/30 samples yielding cumulative sensitivity/specificity values of 64%/100%. PCR performance of single corresponding blood samples of these pts was significantly inferior with only 3/33 patients with proven/probable IPA resulting in a sensitivity/specificity of 8%/87% (p<0.005) GM ELISA positivity in BAL was observed for 22/26 pts with sensitivity of 85%. Identical to PCR testing, GM ELISA was significantly more sensitive for detection of IPA when performed in BAL compared to same-day serum samples (p<0.005).

Conclusions
In this retrospective, direct comparison of clinical specimens we observed a significantly superior performance of an Aspergillus PCR as well as GM ELISA assay directly from the site of infection compared to concurrent blood sample testing for pts with suspected IPA during AFT.
Thus, PCR analyses or GM ELISA for early diagnosis of IA should be preferably directed at specimens such as BAL when IPA is suspected.