(1, 3)-β-D-glucan assay for diagnosing invasive fungal infections in critically ill patients with hematological malignancies

// Elie Azoulay 1 , Nicolas Guigue 1 , Michael Darmon 2 , Djamel Mokart 3 , Virginie Lemiale 1 , Achille Kouatchet 4 , Julien Mayaux 5 , François Vincent 6 , Martine Nyunga 1 , Fabrice Bruneel 1 , Antoine Rabbat 7 , Stéphane Bretagne 1 , Christine Lebert 8 , Anne-Pascale Meert 9 , Dominique Benoit 10 , Frédéric Pene 7 1 Medical ICU and Mycology Department, Saint-Louis Hospital, Paris, France 2 Medical-Surgical ICU, Saint-Etienne University Hospital, Saint-Étienne, France 3 Medical-Surgical ICU Paoli Calmette Cancer Institute, Marseille, France 4 Medical ICU, Angers University Hospital, Angers, France 5 Medical ICU CHU Pitié-Salpêtrière, Paris, France 6 Medical-Surgical ICU, Montfermeil Hospital, Montfermeil, France 7 Medical ICU and Pulmonary Department, Cochin Hospital, Paris, France 8 Medical-Surgical ICU, La Roche Sur Yon Hospital, La Roche Sur Yon, France 9 Medical-Surgical ICU, Jules Bordet Cancer Institute, Brussels, Belgium 10 Medical-Surgical ICU, Ghent University Hospital, Gent, Belgium Correspondence to: Elie Azoulay, e-mail: elie.azoulay@sls.aphp.fr Keywords: invasive fungal infection, intensive care units, diagnostic tests, (1–3)-beta-D-glucan assay Received: August 24, 2015 Accepted: January 19, 2016 Published: February 18, 2016 ABSTRACT Invasive fungal infections (IFIs) are life-threatening complications of hematological malignancies that must be diagnosed early to allow effective treatment. Few data are available on the performance of serum (1–3)-β-D-glucan (BG) assays for diagnosing IFI in patients with hematological malignancies admitted to the intensive care unit (ICU). In this study, 737 consecutive patients with hematological malignancies admitted to 17 ICUs routinely underwent a BG assay at ICU admission. IFIs were diagnosed using standard criteria applied by three independent specialists. Among the 737 patients, 439 (60%) required mechanical ventilation and 273 (37%) died before hospital discharge. Factors known to alter BG concentrations were identified in most patients. IFIs were documented in 78 (10.6%) patients (invasive pulmonary aspergillosis, n = 54; Pneumocystis jirovecii pneumonia, n = 13; candidemia, n = 13; and fusarium infections, n = 3). BG concentrations (pg/mL) were higher in patients with than without IFI (144 (77–510) vs. 50 (30–125), < 0.0001). With 80 pg/mL as the cutoff, sensitivity was 72%, specificity 65%, and area-under-the-curve 0.74 (0.68–0.79). Assuming a prevalence of 10%, the negative and positive predictive values were 94% and 21%. By multivariable analysis, factors independently associated with BG > 80 pg/mL were IFI, admission SOFA score, autologous bone-marrow or hematopoietic stem-cell transplantation, and microbiologically documented bacterial infection. In conclusion, in unselected critically ill hematology patients with factors known to affect serum BG, this biomarker showed only moderate diagnostic performance and rarely detected IFI. However, the negative predictive value was high. Studies are needed to assess whether a negative BG test indicates that antifungal de-escalation is safe.