CROI 2025 Abstract eBook

Abstract eBook

Poster Abstracts

clustering model’s overall performance by analyzing its precision (true positive/ (true positive + false positive)), recall (true positive/(true positive + false negative)), and F-1 score ((2*precision*recall)/(precision +recall)). Results: Of 165,541 deaths recorded in NAP, 163,251 (99%) had specific information on cause of death. A total of 96,350 deaths (58%) were AIDS related, 59,871 (36%) were non-AIDS-related, and 9,320 (6%) were classified as ill-defined. After deduplication, 26,913 unique causes of death and 31 major disease groups were identified. The text clustering model achieved 95% precision, 95% recall, and 95% f-1 score. Precision scores revealed low accuracy for AIDS-related causes from immune deficiency or Lost to Follow-up (79%), hematologic diseases (77%), and assault (52%). Recall scores were low for hematologic disease (24%) and poison (53%). F-1 scores were for low for hematologic disease (37%) and assault (66%). Tuberculosis and cancer had the highest precision (100%), recall (100%), and f1-score (100%) in the model. Conclusions: The text clustering model showed high performance overall, but certain categories had low precision, recall, and F1-scores, possibly due to incomplete information on death certificates. Future improvements to utilize AI to enhance text clustering capabilities may increase system effectiveness. 1167 Changes Over Calendar Time in Discrimination of Prognostic Models for People With HIV Starting ART Martina Cusinato 1 , Suzanne M. Ingle 1 , Adam Trickey 1 , Linda Wittkop 2 , Ard van Sighem 3 , Robert Zangerle 4 , Enrico Girardi 5 , Matthias Cavassini 6 , Bernadino Roca 7 , Amy Justice 8 , Sophie Grabar 9 , Niels Obel 10 , Inma Jarrin 11 , Andreu Bruguera 12 , Jonathan A. C. Sterne 1 , for the Antiretroviral Therapy Cohort Collaboration (ART-CC) 1 University of Bristol, Bristol, UK, 2 University of Bordeaux, Bordeaux, France, 3 Stichting HIV Monitoring, Amsterdam, Netherlands, 4 Medical University of Innsbruck, Innsbruck, Austria, 5 IRCCS Lazzaro Spallanzani, Rome, Italy, 6 Lausanne University Hospital, Lausanne, Switzerland, 7 Hospital General Universitari de Castelló, Castelló, Spain, 8 VA Connecticut Healthcare System, West Haven, CT, USA, 9 Sorbonne University, Paris, France, 10 Copenhagen University Hospital, Copenhagen, Denmark, 11 Instituto de Salud Carlos III, Madrid, Spain, 12 Centre d'Estudis Epidemiològics Sobre les ITS i Sida de Catalunya, Barcelona, Spain Background: Accurately predicting survival of persons with HIV (PWH) is important to guide clinical decision-making and lifestyle choices. The Antiretroviral Therapy Cohort Collaboration (ART-CC) developed prognostic models estimating three- and five-year mortality risk among PWH starting ART between 1995-2003. Using more recent data to re-fit these models, we investigated changes in discrimination to predicting mortality for more recent time periods. Methods: Data were combined from 18 cohort studies in 9 countries in Western Europe and North America. Eligible PWH were ART-naïve adults (≥16 years) starting treatment with ≥3 antiretroviral drugs from 1996-2019. PWH were followed from ART start date until death (outcome variable) or censoring. Three- and five-year follow-up periods were investigated. Model covariates were sex, age, HIV acquisition group, HIV-1 RNA, CD4 count, and AIDS/no AIDS before ART start. C-statistics for discrimination were derived from adjusted Weibull models for ART start year periods 1996-2001, 2002-2006, 2007-2009, 2010-2013 and 2014-2019. Results: Among 159,257 PWH starting ART, 6755 died within five years (683,000 person-years) and 4,987 within three years (319,000 person-years). The proportion of women was stable overtime, while the proportion of participants starting ART aged ≥50 years and with higher CD4 counts increased. Median follow-up was five years except for PWH starting ART between 2014 2019 (4.3 years). Three-year adjusted mortality hazard ratios (aHRs) were similar to those for five years shown in the table. The aHRs for prior AIDS (versus no AIDS) increased from 2.59 (95% confidence interval: 2.32-2.88) for ART start year 1996-2001 to 3.80 (3.27-4.41) for 2014-2019. The aHRs for IDU (versus MSM) also increased over time, from 2.92 (2.57-3.33) to 4.09 (3.30-5.07). The aHRs for all age groups (versus 16-29) increased, whilst the aHRs for CD4 categories 0-99 and 100-199 (versus 200-349) decreased. C-statistics for the 5-year models increased from 0.769 to 0.834 between 1996-2001 and 2014-2019, whilst 3-year model C-statistics increased from 0.781 to 0.845. Conclusions: Discrimination of prognostic models based on standard markers for mortality among PWH starting ART has improved since combination ART became widely available in western Europe and North America. Risk profiles of PWH starting ART changed markedly due to earlier diagnosis and treatment. Future work will assess the impact on discrimination of additional predictors.

1168 Clinical Outcomes in Lung Transplantation in People Living With HIV: A Competing Risk Analysis German Contreras, George Golovko University of Texas Medical Branch, Galveston, TX, USA Background: Little is known about post-transplant outcomes in people living with HIV (PLWH). Traditional survival analysis typically assumes the presence of a single outcome, but in real-world scenarios, multiple competing events can influence survival and other key metrics. This study aims to assess clinical outcomes in PLWH post-lung transplantation (LTx) using a competing risk analysis to better understand the true incidence of complications and death. Methods: We conducted a retrospective cohort study of PLWH and non-PLWH, aged 18 and older, who underwent LTx between 2022 and 2024. Data were collected from 64 healthcare organizations within the TriNetX U.S. Network. We applied the Aalen-Johansen method to estimate the cumulative incidence function (CIF) for three post-transplant outcomes: non-infection complications (surgical complications), infection complications (CMV, bacterial pneumonia, bacteremia, mycosis, CDI, RSV), and death. The analysis spanned one year after transplantation, stratifying by HIV status. Results: A total of 63 PLWH and 3,874 non-PLWH were included in the study. The median age at the time of transplantation was 56 years for both groups, with a higher proportion of males (66.1% vs. 59%) and non-Hispanics (71% vs. 67.1%) in each respective cohort. Non-infection complications were the most common outcome in both groups, with similar CIFs (35.3% vs. 32.5%) Table. PLWH had a higher CIF for infection-related complications (25.3% vs. 18.7%) yet a lower mortality CIF (3.8% vs. 6.9%) than non-PLWH Table. CMV infection was the most frequent infection complication in both groups, with PLWH experiencing a higher incidence within the first year (18.1% vs. 13.6%) Table. Notably, the risk of mycosis was higher in PLWH (5.5% vs. 2.6%), while RSV cumulative incidence was significantly higher in the non-PLWH group Table. Conclusions: PLWH undergoing LTx experience a distinct post-transplant complication profile, particularly regarding infection risks. Despite a higher incidence of infection complications, PLWH had lower mortality within the first year post-LTx, suggesting that further investigation into tailored infection prevention strategies and post-transplant management is warranted. These findings underscore the importance of using competing risk analysis to capture the full spectrum of outcomes in this unique patient population.

Poster Abstracts

CROI 2025 382