CROI 2020 Abstract eBook

Abstract eBook

Poster Abstracts

reported the accuracy and predictive values of the Mp1p EIA in a prospective study Methods: We consecutively recruited HIV patients aged ≥18 who were hospitalized at the Hospital for Tropical Diseases in Ho Chi Minh city with any symptoms, had a CD4 count <100 cells/mL, were ART-naïve, or were on ART for <3 months or >12 months. Serum, plasma, and urine samples were collected for Mp1p testing alongside blood cultures. All patients were followed over 6 months. Diagnostic accuracy and predictive values were calculated based on the reference standard, defined as cumulative incidence of culture-positive Tm from any sterile sites over 6 months. Results: 533 patients were recruited between January 2018 and July 2019. 78.4%were male; median age was 34 (IQR: 29-40) years; median CD4 count was 16 (IQR: 6-36) cells/mL. A total of 81 (15.2%) patients developed Tm: 69 during hospitalization, 12 during the follow up period, with sensitivity of cultures of 85.2%. The AUCs generated from the ROC curves in sera, plasma, and urine were 98.1%, 96.2%, 97.2%, respectively. Based on the OD cut off generated by the Youden indexes of 0.17, 0.23, and 0.39 for sera, plasma, and urine, respectively, the sensitivity was 88.9%, 90.3%, 94.4%; specificity was 96.6%, 96.8%, 98.8% (Psera-urine=0.011, Pplasma-urine=0.013, McNemar test); PPV 83.2%, 85.1%, 98.8%; NPV 98.2%, 98.4%, 99.0%. When testing all 3 specimens in combination, sensitivity=97.5%, specificity=96.5%, PPV=82.8%, NPV=99.1%. The sensitivity when testing all 3 specimens was higher than cultures (97.5% vs 85.2%, P<0.001, McNemar test). In 12 patients, Tm antigen was positive 1 to 16 weeks before cultures turned positive. Conclusion: The Mp1p EIA offers superior sensitivity and excellent specificity (>95%) compared to cultures in diagnosing Tm. Urine is as sensitive and is more specific than plasma and sera for antigen testing. This test can detect Tm up to 4 months before cultures turn positive and could transform the current management of talaromycosis. 751 IN VITRO ANTIFUNGAL SUSCEPTIBILITY AND ANTIFUNGAL TREATMENT OUTCOME IN TALAROMYCOSIS Thu T. Nguyen 1 , Hannah Shepard 2 , Ly T. Vo 3 , Thanh T. Nguyen 4 , Thuy Le 1 1 Duke University School of Medicine, Durham, NC, USA, 2 Duke University, Durham, NC, USA, 3 Ho Chi Minh City Medicine and Pharmacy University, Hồ Chí Minh, Vietnam, 4 Oxford University Clinical Research Unit in Vietnam, Ho Chi Minh, Vietnam Background: The dimorphic fungus Talaromyces marneffei (Tm) causes an invasive mycosis which ranks 3rd as the most common HIV-associated infections in Southeast Asia with a mortality as high as 30%. We recently demonstrated in a randomized control trial (N=440 patients) that induction therapy with itraconazole was associated with higher mortality, persistent fungemia, incidence of relapse and IRIS when compared to amphotericin B over six months. We hypothesize that disease relapse and other complications in patients who received itraconazole are associated with a reduced susceptibility to itraconazole. Currently methods for antifungal susceptibility testing (AFST) and clinical breakpoints to define antifungal resistance have not been established for Tm. Methods: To test our hypothesis, we developed a new AFST testing method for Tm. We followed the CLSI guidelines for broth microdilution of itraconazole and preparation of a standardized yeast inoculum of 10 6 cells/ml. We utilized alamarBlue, a dye which fluoresces as a result of cellular metabolic activity, allowing percent reduction in fluorescence intensity to be precisely calculated. We generated MIC 50 and MIC 90 values for 136 unique Tm strains isolated from patients treated with itraconazole, and we compared the MIC geometric means in patients who had a good treatment outcome and multiple groups of patients who had poor treatment outcomes. Results: The assay performed consistently with intra-assay MICs of 0.008μg/ mL for 6 sample replicates, and inter-assay MICs testing 6 runs on separate

days were within the CLSI acceptable range of one 2-fold dilution. Among 136 isolates, 79% had MIC 90 =0.008 μg/mL, 16% had MIC 90 =0.016, and 4% had MIC 90 =0.03. In multiple pairwise comparisons, the differences in MIC 90 geometric means between patients who responded well to itraconazole (N=59) and patients who had any bad outcome (N=77), including death (N=23), relapse (N=9), prolonged fungemia (N=55), and IRIS (N=14) were not statistically significant, all P values fromWilcoxson rank sum tests were >0.05. Conclusion: We developed a highly reliable and reproducible method for in-vitro AFST for Tm in the yeast form. The use of alamarBlue enables precise quantification of MIC without relying on visual perception and can be standardized across laboratories. The MICs against itraconazole in all isolates were low (≤0.03 μg/mL), and the MIC distribution did not correlate with the outcome of itraconazole therapy in HIV-associated talaromycosis.

Poster Abstracts

752 PREVALENCE OF CMV VIREMIA AND ASSOCIATED RISK IN HIV-INFECTED PERSONS STARTING ART Shweta Sharma 1 , Mark R. Schleiss 1 , Hansjakob Furrer 2 , Daniel Nixon 3 , Mark Blackstad 1 , Nelmary Hernandez-Alvarado 1 , Dominic Dwyer 4 , Álvaro H. Borges 5 , H. Clifford Lane 6 , Jean-Michel Molina 7 , Jens D. Lundgren 5 , James Neaton 1 , for the INSIGHT and ANRS groups. 1 University of Minnesota, Minneapolis, MN, USA, 2 University Hospital of Bern, Bern, Switzerland, 3 Virginia Commonwealth University, Richmond, VA, USA, 4 Westmead Hospital, Westmead, Australia, 5 CHIP, Department of Infectious Diseases, Copenhagen, Denmark, NIAID, Bethesda, MD, USA, 7 Hôpital Saint-Louis, Paris, France Background: Morbidity and mortality in advanced HIV-infection is still high despite ART use and prophylaxis for opportunistic infections. Data on prevalence of CMV viremia pre- and post- ART initiation at varying CD4 thresholds are limited. The impact of CMV viremia on morbidity and mortality is unclear. Methods: Using plasma samples from participants initiating ART at study entry in 4 clinical trials (INSIGHT: FIRST, SMART, START, and ANRS: REFLATE), we measured CMV-specific IgG and CMV viremia at baseline and in year 1 visits. CMV DNA was measured centrally. Detectable (lower limit 88.5 IU/mL) CMV DNA was used for CMV+/CMV- classification. CMV+ during follow-up was defined as CMV DNA at any visit. Analyses for the association of CMV+/CMV- with clinical risk were limited to FIRST (longer follow-up and number of outcomes). Using all follow-up in FIRST, we estimated the hazard ratio (HR) for baseline CMV+ vs. CMV- for a composite outcome of AIDS, serious non-AIDS (SNA), or death. HRs were also computed for outcomes after 8 months of ART across 4 subgroups defined by baseline and follow-up CMV+/CMV- through month 8. Models were adjusted for CD4 counts and HIV RNA. Results: There were 1169 participants from FIRST, 137 from REFLATE, 54 from SMART, and 1815 from START with median baseline CD4 counts of 153, 140, 429, and 648 cells/µL, respectively. CMV infection by IgG was ≥ 90% across trials. Baseline CMV+ prevalence was 17%, 26%, 1%, and 0% in FIRST, REFLATE, START, and SMART, respectively. Pooled across trials, baseline CMV+ prevalence by CD4 count was: ≤50 cells/µL: 39% of 382; 51-100: 29% of 148; 101-200: 12% of 221; 201-350: 4% of 289; 351-500: 1% of 195; > 500: 1% of 1941. FIRST participants were grouped using baseline and follow-up CMV+/CMV- at months 4 and 8 (n=1151): 2%were baseline CMV- but follow-up CMV+, 3%were baseline and follow-up CMV+, and 14%were baseline CMV+ and follow-up CMV-. Using baseline, months 4 and 6 in REFLATE (n=128), these percentages were 5%, 2%, and 21%. START and SMART percentages were <1. Table 1 presents HRs for the composite outcome across subgroups defined by baseline and/or follow-up CMV+/CMV- in FIRST. Conclusion: Prevalence of CMV viremia at baseline was higher among ART- naïve HIV+ persons with lower CD4 counts. Persistent or development of CMV

CROI 2020 277