S10.2d Fungal beta-glucans and mannan performances in HIV-associated histoplasmosis

Abstract S10.2 Fungal Infections in Transplant Patients, September 24, 2022, 10:30 AM - 12:00 PM Objectives Diagnosis of histoplasmosis in people living with HIV (PLWHIV) remains challenging despite developments in Histoplasma antigen and molecular detection tools. Fungal markers such as Beta-(1,3)-D-glucan (BDG) and galactomannan Aspergillus antigen (GM) are widely available, but the experience is limited during PLWHIV workup for suspicion of histoplasmosis. Our objective was to evaluate and compare BDG and GM performances for the diagnosis of HIV-associated histoplasmosis. Methods We performed a diagnostic accuracy study using primary serum samples stored frozen in a certified biorepository (CRB Amazonie-DC-2021-4649). Samples consisted of consecutive hospitalized PLWHIV unexposed to oral antifungals during the previous month (EDIRAPHIS study-NCT01884779). All patients gave consent for biobanking and ancillary studies on fungal markers. Histoplasmosis cases, proven (EORTC/MSG criteria) and probable (polyclonal or monoclonal Histoplasma antigen detections in urines or serum), as well as negative controls, were randomly selected. Patients with a proven or suspected Pneumocystis jirovecii infection were excluded. Following manufacturers’ instructions, samples were blindly tested for BDG and GM using Fungitell® and PlateliaTM Aspergillus Ag assays, respectively. Gold standard definition used three scenarios: EORTC scenario (cases and controls defined according to the EORTC/MSG 2020 criteria for endemic mycoses); strict scenario with proven cases restricted to those positives to all three previous Histoplasma antigen detections, and controls conversely negatives for all methods; and a large scenario with proven or probable cases and controls negatives for all methods. Results We included 121 samples, 92 HIV-associated histoplasmosis cases (34 proven and 58 probable), and 29 negative controls. Compared with controls, histoplasmosis cases were significantly younger and advanced in the course of HIV disease [median CD4 count level 33/mm3 (15-87) vs 116 (62-245)]. BDG and GM median detection levels were significantly higher among histoplasmosis cases compared with controls across all scenarios [368 (176-441) pg/ml vs 142 (89-211) for BDG and 2.5 (1.3-4.8) vs 0.19 (0.14-0.36) for GM, in cases vs controls of the strict scenario, respectively]. In the strict scenario, at 150 pg/ml and 0.5 for BDG and GM respectively, sensitivity, specificity, positive and negative likelihood ratios were respectively: 95% [95%confidence interval (CI), 85-100] vs 90% [77-100], 52% [34-70] vs 83% [69-97], 2 [1.4-3.0] vs 5.3 [2.4-12.0], and 0.1 [0.01-0.7] vs 0.12 [0.03-0.45]. ROC curves found AUCs of 0.82 [0.68-0.91] vs 0.92 [0.80-0.98], and optimal thresholds at 288 pg/ml and 1.29, for BDG and GM, respectively (Fig. 1). Post-test probabilities showed best performances at the lowest thresholds for negative testing of both BDG and GM, and at the 0.7 thresholds for a positive GM test (Fig. 2). Conclusion BDG and GM may not be used for the same objective when searching for HIV-associated histoplasmosis. Although a negative BDG test at the lowest thresholds should rule out histoplasmosis in a screening context, limitations of a positive BDG test, even at the highest thresholds, call for a consecutive positive GM test before starting patients on antifungal therapy targeting histoplasmosis. Still, when considering the highest costs of BDG testing, higher balanced diagnostic performances, and lower costs of GM testing alone, one may favor the use of GM, notably in resources-limited settings.