CROI 2019 Abstract eBook

Abstract eBook

Poster Abstracts

Tatiana Caceres 6 , Eduardo Gotuzzo 6 , Charles Cooper 2 , Susan E. Dorman 7 , Yukari C. Manabe 1 1 Johns Hopkins University School of Medicine, Baltimore, MD, USA, 2 Becton Dickinson, Franklin Lakes, NJ, USA, 3 University of Cape Town, Cape Town, South Africa, 4 Infectious Disease Institute, Kampala, Uganda, 5 Byramjee Jeejeebhoy Government Medical College, Pune, India, 6 Universidad Peruana Cayetano Heredia, Lima, Peru, 7 Medical University of South Carolina, Charleston, SC, USA Background: Tuberculosis (TB) control among people living with HIV requires accurate, rapid diagnostic tests to identify Mycobacterium tuberculosis complex (MTBc), and detect isoniazid (INH) and rifampin (RIF) resistance. We evaluated the performance of the BD MAX™ MDR-TB (BD MAX) assay, which tests up to 24 specimens at once. Methods: Outpatient adults with signs and/or symptoms of active pulmonary TB were enrolled in South Africa, Uganda, India, and Peru. A single collected sputumwas split into 2 portions. Smear microscopy and BD MAX were performed on the raw portion. The other portion was processed using NALC- NaOH, and tested using culture (MGIT™), phenotypic drug susceptibility testing, Xpert® MTB/RIF (Xpert), BD MAX, and microscopy. Results: 1053 participants (47% female, 32% HIV-infected) with presumptive TB were enrolled. The majority (94%) of HIV-infected participants were enrolled from high HIV/TB burden sites (Uganda and South Africa) with median CD4 of 367 (IQR 228-536). Overall BD MAX test sensitivity was 93% (262/282 [95% CI 89,95]) among microbiologically confirmed TB participants, and specificity was 97% (593/610, [96,98]) among participants with negative cultures. Among 273 HIV-infected participants, sensitivity was 86% (44/51 [74,93]), and specificity was 98% (217/222 [95,99]) When stratified by ZN smear microscopy status, sensitivity of BD MAX for detection of HIV-associated TB was 100% (22/22, [85, 100]) for smear-positive, and 76% (22/29, [58, 88]) in smear-negative HIV/TB patients. Among TB patients with both BD MAX and Xpert results, sensitivity was 82% (41/50, [69, 90]) for both assays. BD MAX sensitivity for detection of any drug resistance (INH and/or RIF) was 100% (4/4 [51,100]), with specificity among those with drug-susceptible TB of 100% (30/30 [89,100]) when compared to MGIT DST among HIV/TB participants. Sensitivity and specificity for detection of INH resistance was 100% (3/3 [44, 100]) and 100% (35/35, [90, 100]), respectively. Sensitivity for RIF resistance was 100% (2/2, [34,100]). One participant had RIF resistance by BD MAX, but was considered susceptible on MGIT DST and bi-directional sequencing, resulting in an overall specificity of 97% (34/35, [85,99]). Conclusion: The BD MAX MDR-TB assay has high sensitivity and specificity for detection of MTBc, and rifampin and isoniazid drug resistance in settings of high HIV/TB burden and may aid in the rapid detection of tuberculosis and MDR-TB. 720 THE EFFECTIVENESS OF VARIOUS SYMPTOM-BASED ALGORITHMS FOR TB SCREENING AT HIV TESTING Vanessa Rivera 1 , Marc Antoine Jean Juste 1 , Jeanwilkens Sainristil 1 , Dani Archange 1 , Samantha Gluck 2 , Harrison Reeder 2 , Oksana Ocheretina 3 , Tahera Doctor 2 , Eleanore Fuqua 2 , Sarah Centanni 2 , Pierre Cremieux 2 , Daniel Fitzgerald 3 , Serena Koenig 4 , Jean William Pape 1 1 GHESKIO, Port-au-Prince, Haiti, 2 Analysis Group, Inc, Boston, MA, USA, 3 Weill Cornell Medicine, New York, NY, USA, 4 Brigham and Women’s Hospital, Boston, MA, USA Background: It is essential to screen for TB prior to ART initiation in TB endemic settings. Algorithms are needed to identify low-risk patients for immediate ART initiation, and high-risk patients who merit TB testing prior to ART initiation. If TB screening is conducted at the time of HIV testing, symptomatic patients may also be tested for TB, even if they test negative for HIV. We conducted a retrospective analysis to evaluate the diagnostic yield of five different symptom screening strategies among patients who presented for HIV testing in Haiti. Methods: From October 1, 2015 to March 31, 2016, the first 20 patients who presented for HIV testing each day at GHESKIO were queried regarding TB symptoms, and received AFB smear, GeneXpert testing, and chest x-ray, regardless of symptoms. TB symptom screening algorithms were evaluated based on the total proportion of TB cases diagnosed and the number of missed TB cases, stratified by HIV status. Results: 1,108 individuals received diagnostic testing for both HIV and TB. 48% were female, median age was 33 (27-44) and 216 (19%) tested positive for HIV, with median CD4 count of 364 cells/mm3 (IQR: 210-563). 59 patients (5%) were diagnosed with TB; 55 (93%) of these were bacteriologically confirmed, and 4 (7%) were diagnosed based on symptoms and chest x-ray. Among the 216 persons who tested positive for HIV, 13 (21%) of those who reported cough of

any duration were diagnosed with TB (Table 1). Of non-coughing patients with HIV, 2 (1%) were diagnosed with TB; these patients also did not report fever, night sweats, or weight loss. Among the 892 patients who tested negative for HIV, 36 (16%) of those who reported cough of any duration were diagnosed with TB. Among HIV-negative patients without cough, 8 (1%) were diagnosed with TB; 3 of these patients reported fever, night sweats or weight loss in the absence of cough. Conclusion: Testing for TB in patients who do not report cough at HIV testing is low yield in Haiti, regardless of HIV test results; ART can generally be initiated in patients who do not report cough without further TB testing. Patients who report cough of any duration at HIV testing should be evaluated for TB, regardless of HIV test results. Patients newly diagnosed with HIV should be screened for cough of any duration, and those reporting cough should be tested for TB prior to ART initiation.

Poster Abstracts

721 OPTIMIZING DIAGNOSTIC ALGORITHMS FOR ACTIVE PULMONARY TUBERCULOSIS IN HIV CLINICS Haylea A. Hannah 1 , Bradley G. Wagner 2 , Stewart T. Chang 2 , Paul K. Drain 1 1 University of Washington, Seattle, WA, USA, 2 Institute for Disease Modeling, Bellevue, WA, USA Background: Most TB deaths are preventable with early diagnosis and treatment among people living with HIV (PLHIV). Current testing algorithms rely on symptom screening and sputum-based diagnostic tests, which are less sensitive among PLHIV and often result in treatment delays and loss to follow- up; nearly 20% of TB cases go undiagnosed in South Africa. Point-of-care (POC) testing in HIV clinics may improve case-detection, thereby reducing TB disease incidence and mortality. Methods: We used EMOD-TB, an individual-based TB transmission model, to estimate the impact of HIV clinic-based screening-diagnostic algorithms on rates of TB disease incidence and mortality in South Africa. The model accounted for the natural history of TB and HIV, disease progression, the TB care cascade, and historical estimates of country-level disease incidence and mortality. The model assumes each algorithm is offered to all clinic attendees beginning in 2016 when receiving HIV testing and annually thereafter. Test sensitivities and specificities differed by HIV status and CD4 count. The five algorithms, with each test based on a preceding positive result, were: 1) Four TB symptom screening (4SS)+GeneXpert (Xpert); 2) 4SS+urine lipoarabinomannan (uLAM)+Xpert (to reduce testing costs); 3) 4SS+uLAM; 4) C-reactive protein (CRP)+uLAM+Xpert; and 5) CRP+uLAM. Results: Incorporating intensified TB case finding into routine HIV testing resulted in a reduction in predicted TB incidence and mortality. The algorithm of 4SS+Xpert yielded the greatest overall decline in TB burden with an estimated additional reduction of 12% (11-13%) in incidence and 16% in mortality (15-17%) from 2016 to 2025 compared to baseline trends. Both 4SS+uLAM and CRP+uLAM+Xpert resulted in reductions of an additional 4% in TB incidence (3-5%) and 6% in TB mortality (7-10%) compared to baseline trends. Administering uLAM before confirmatory testing resulted in a 90% reduction in the number of Xpert tests ordered in the HIV clinic. Conclusion: Incorporating 4SS+Xpert testing into HIV care would significantly reduce TB disease incidence and mortality in South Africa. We predict that HIV clinic-based testing algorithms that incorporate CRP and uLAM would result in smaller declines in burden, however uLAM testing would have a greater relative impact on mortality among PLHIV with lower CD4 counts. HIV clinic-based testing algorithms that incorporate CRP and uLAM should be evaluated in prospective clinical trials with respect to improving TB outcomes among PLHIV.

CROI 2019 277