CROI 2025 Abstract eBook

Abstract eBook

Poster Abstracts

961

Machine Learning-Based Prediction of Active Tuberculosis in People With HIV Using Clinical Data Lena Bartl 1 , Marius Zeeb 2 , Marisa Kälin 2 , Tom Loosli 2 , Matthias Hoffmann 3 , Katharina Grabmeier 4 , Michael Knappik 5 , Alexandra Calmy 6 , Jose Damas 7 , Niklaus D. Labhardt 8 , Huldrych F. Günthard 2 , Roger D. Kouyos 2 , Katharina Kusejko 2 , Johannes Nemeth 1 , for the Swiss HIV Cohort Study 1 University Hospital Zurich, Zurich, Switzerland, 2 University of Zurich, Zurich, Switzerland, 3 Bern University Hospital, Bern, Switzerland, 4 Medical University of Vienna, Vienna, Austria, 5 Otto-Wagner Hospital, Vienna, Austria, 6 University Hospitals of Geneva, Geneva, Switzerland, 7 Centre Hospitalier Universitaire Vaudois, Lausanne, Switzerland, 8 University of Basel, Basel, Switzerland Background: Coinfections of Mycobacterium tuberculosis (MTB) and human immunodeficiency virus (HIV) represent a significant global health challenge. Patients infected with MTB are at elevated risk of progressing to active tuberculosis (TB), a risk that can be mitigated by preventive therapy. However, current testing methods frequently fail to identify individuals who will later develop active TB, particularly among people with HIV (PWH). Methods: We developed random forest models to predict progression to active TB using clinical data collected at the time of HIV-1 diagnosis. The model was developed using a training and test set approach within the Swiss HIV Cohort Study (SHCS), which included 55 PWH who developed active TB within six months of enrollment and 1432 matched controls without active TB, enrolled between 2000 and 2023. After internal validation within SHCS, external validation was performed using data from the Austrian HIV Cohort Study (AHIVCOS), comprising 43 incident TB cases and 1005 controls. Results: The model predicted active TB with an area under the receiver operating characteristic (ROC) curve (AUC) of 0.83 (95% CI 0.8-0.86) in the SHCS. After adjusting for demographic variables and re-fitting the model with fewer parameters, the AUC was 0.72 for Swiss PWH and 0.67 for Austrian PWH. Demographic parameters, particularly region of origin and ethnicity, significantly influenced model performance, likely due to varying TB incidence rates. Socioeconomic factors, such as profession and education, along with mode of HIV transmission, also impacted the model’s predictions. Additionally, laboratory metrics such as CD4 cell count and HIV RNA levels, which reflect immune system status, were critical in predicting TB progression. Overall health markers, such as BMI, creatinine, and hemoglobin, also affected model accuracy. The model outperformed standard diagnostic tools (tuberculin skin test and interferon-gamma release assay) in identifying high-risk individuals, demonstrated by a lower number needed to diagnose (1.96 vs. 4). Conclusions: Machine learning-based models show considerable potential to improve care for PWH, as they require no additional data collection, incur minimal costs, and enhance the identification of those who could benefit from preventive TB treatment.

962

Genome-Wide Meta-Analysis Identifies Genetic Associations With Resistance to Mtb Infection Matheus Fernandes Gyorfy 1 , Neel R. Gandhi 1 , Qin Hui 1 , Bruno B. Andrade 2 , Neil Martinson 3 , Mandar Paradkar 4 , Senbagavalli Prakash 5 , Kamakshi Prudhula Devalraju 6 , Marina C. Figueiredo 7 , Marcelo Cordeiro-Santos 8 , Nombuyiselo Mofokeng 3 , Fay Willis 1 , Timothy Sterling 7 , Amita Gupta 9 , Yan V. Sun 1 , for the TB GWAS Collaboration 1 Emory University, Atlanta, GA, USA, 2 Instituto Gonçalo Moniz, Salvador, Brazil, 3 Perinatal HIV Research Unit, Soweto, South Africa, 4 The Johns Hopkins Center for Infectious Diseases in India (CIDI), Pune, India, 5 Jawaharlal Institute of Postgraduate Medical Education and Research, Puducherry, India, 6 Bhagwan Mahavir Medical Research Centre, Hyderabad, India, 7 Vanderbilt University Medical Center, Nashville, TN, USA, 8 Fundação de Medicina Tropical Doutor Heitor Vieira Dourado, Manaus, Brazil, 9 The Johns Hopkins University School of Medicine, Baltimore, MD, USA Background: Tuberculosis (TB) is the leading infectious cause of death worldwide, including among people living with HIV. The WHO End TB goal of reducing TB incidence to <10 cases per 100,000 will not be achieved without a greater understanding of host mechanisms that mediate susceptibility to Mycobacterium tuberculosis (Mtb) . A genetic basis for “resistance” to Mtb infection has been postulated, but the relationship between host genetics and Mtb infection is underexplored. We conducted a genome-wide association study (GWAS) to identify loci associated with Mtb resistance and infection among close contacts of TB patients in Brazil, India, and South Africa. Methods: Among 4,370 close contacts of microbiologically-confirmed, active pulmonary TB patients, we categorized contacts as “resisters” if they had: high exposure (slept in the same room or >4 hours indoors), to a highly infectious index case (cavitary or high smear grade), and negative tuberculin skin tests (TST) and/or interferon-gamma release assays (IGRA). Contacts with positive TST and/or IGRA were categorized as Mtb infected. After quality control at sample and variant levels, we conducted country-specific GWAS and meta-analysis of the resister and infected phenotypes, adjusted for age, sex, population structure, and relatedness using TOPMed-imputed data. Results: Among the contacts enrolled, we identified 457 individuals with resistance to Mtb infection, 1,903 with positive TST and/or IGRA; 361 with discordant or missing TST/IGRA results, and 1,168 with negative TST/IGRA but low or medium exposure to the TB index patient. In the meta-analysis for resistance to Mtb infection, we identified two genome-wide significant (GWS) loci associated with resistance: a chromosome 13 locus, close to MYO16 gene (rs1295104126, OR = 0.92 [95%CI: 0.89, 0.94]; p = 1.60x10 -10 ), and a chromosome 2 locus, mapped to PARD3B gene (rs888091, OR = 0.83 [95%CI: 0.78, 0.89]; p = 4.64x10 -9 ). In the meta-analysis for Mtb infection, we identified one GWS locus close to HLA-DQA1 gene on chromosome 6 (rs28752534, OR = 0.92 [95%CI: 0.89, 0.94], p = 6.58x10 -12 ). Conclusions: Our robust methods uncovered multiple statistically significant associations. Further investigation into the functional roles of these genetic loci is warranted. Consistent results from diverse ancestral groups may help us better understand the molecular mechanisms underlying Mtb resistance and infection, thus improving the generalizability of our findings across global populations with high TB burden. Anura David 1 , Gregory P. Bisson 2 , Salome Charalambous 3 , Lesley E. Scott 1 , Griffiths Kubeka 3 , Lyndel Singh 1 , Pedro Da Silva 4 , Wendy S. Stevens 4 , Yeonsoo Baik 2 1 University of the Witwatersrand, Johannesburg, South Africa, 2 University of Pennsylvania, Philadelphia, PA, USA, 3 The Aurum Institute, Johannesburg, South Africa, 4 National Health Laboratory Service, Johannesburg, South Africa Background: Most studies have focused on either healthcare worker collected (HCWC) or self-collected (SC) tongue swabs (TS) for detecting Mycobacterium tuberculosis , but rarely on both. We assessed this, on TS collected from the same participant, using the Xpert MTB/RIF Ultra (Ultra) (Cepheid, Sunnyvale, CA, USA) along with participant acceptability of the collection procedure. Methods: In this cross-sectional study, 669 participants who met eligibility were consented and enrolled at eight healthcare facilities in the Ekurhuleni District of Johannesburg, South Africa. Sputum was tested on Ultra and mycobacterial growth indicator tube (MGIT) culture and two TS (HCWC and SC) tested on Ultra. A total of 314 participants used a short-shaft (4 cm) spun polyester XpressCollect™ swab (Steripack), while 355 participants used a longer shaft (15 cm) Copan FLOQ swab (Copan Italia S.p.A.). Sputum and swabs (dry) Healthcare-Worker Versus Self-Collected Tongue Swabs for Mycobacterium tuberculosis Detection

Poster Abstracts

963

CROI 2025 303