BACKGROUND:Data on predictors of treatment failure in children starting antiretroviral therapy (ART) are limited, particularly on dolutegravir-based regimens (DTG). METHODS:ODYSSEY demonstrated superior efficacy of DTG versus standard-of-care (SOC). We assessed predictors at ART initiation of treatment failure by 96 weeks. RESULTS:Three hundred and eighty-one children started first-line ART (82% African). At ART-initiation, median age was 10.5 years (IQR: 6.5, 14.0, 67 < 3 years), CD4% 20% (IQR: 12, 28), BMI-for-age Z-score -.58 (IQR:-1.48, +.25). One hundred and eighty-nine children started DTG, 192 started SOC (91% ≥3 years started efavirenz; 79% <3 years started lopinavir). Seventy-five children experienced treatment failure (24 DTG, 51 SOC). Failure risk was lower on DTG than SOC (hazard ratio [HR] = 0.47, 95% CI: 0.29-0.77, P = .002). Lower BMI-for-age Z-score (HR = 0.82 for each unit gain, 95% CI: 0.70-0.96, P = .01) and being at an African site (HR = 2.09, 95% CI: 0.82-5.31, P = .09) were associated with higher failure risk. Risk was also higher at younger ages with the steepest increase in the youngest children and increased at lower CD4%, with a stronger CD4% effect at younger ages. At CD4% = 20, HRs relative to age 10 years were 2.40 (95% CI: 1.58-3.65) at age 1 year, 1.30 (95% CI: 1.15-1.48) at age 5 years, and 0.80 (95% CI: 0.72-0.89) at age 18 years. At age 1 year, HRs relative to CD4% = 20 were 1.39 (95% CI: 1.16-1.66) at CD4% = 15, and 0.52 (95% CI: 0.36-0.75) at CD4% = 30; at age 10, corresponding estimates were 1.07 (95% CI: 0.94-1.20) at CD4% = 15, and 0.88 (95% CI: 0.69-1.13) at CD4% = 30. CONCLUSIONS:Young age, low BMI-for-age, and low CD4% at ART initiation predicted higher risk of treatment failure and can guide targeted support.
Although substantial progress has been achieved through the scale-up of antiretroviral therapy (ART), the HIV epidemic in the Asia-Pacific region remains characterized by late presentation, ongoing stigma, and structural inequities in access to care. At the same time, people with HIV (PWH) are aging, resulting in a growing burden of multimorbidity and increasingly complex clinical needs. The Bangkok Symposium International Symposium on HIV Medicine convened regional and international experts to review recent scientific advances and implementation challenges shaping the future of HIV prevention and treatment, with particular emphasis on long-acting agents, differentiated care models, and diagnostic and digital innovations.
OBJECTIVE:This study aimed to develop machine learning (ML) models to predict HIV status and assessed the factors associated with HIV infection among young men who have sex with men (MSM) under the Universal Health Coverage (UHC) programme in Thailand. METHODS:Young MSM aged 15-24 years who underwent HIV testing through the UHC programme from 2015 to 2022 were included. Data were divided into training (70%) and testing (30%) sets, with the Synthetic Minority Oversampling Technique (SMOTE) applied to address data set imbalance. ML models, including logistic regression, k-nearest neighbour (KNN), random forest, extreme gradient boosting (XGB) and AdaBoost, were used to predict HIV infection. RESULTS:Among 146 813 young MSM, 11% were diagnosed with HIV. While KNN initially outperformed other ML models, the sensitivity of all models using the original data set was low due to imbalanced data. After applying SMOTE, the XGB model showed the best performance with an accuracy of 0.72, sensitivity of 0.73, specificity of 0.72 and the area under the curve of 0.72. The top predictors of HIV infection were the year of HIV testing (68%), age (55%) and targeted HIV testing (54%). DISCUSSION:This study demonstrates the potential of ML models, particularly XGB, in predicting HIV infection among young MSM in Thailand under the UHC programme. The application of SMOTE improved model sensitivity, addressing data imbalance and enhancing predictive accuracy. CONCLUSIONS:ML models have the potential to enhance HIV risk assessment and inform targeted prevention strategies for high-risk populations.
BACKGROUND:Simplified approaches to hepatitis C virus (HCV) treatment delivery are needed to meet elimination goals. However, the impact of low-touch strategies on individuals at higher risk due to treatment failure or reinfection is unknown. We estimated HCV reinfection rates, and the impact of resistance associated substitutions (RASs) on response in the ACTG A5360 (MINMON) trial. METHODS:HCV RNA evaluations were scheduled at weeks 0, 24 (sustained viral response [SVR] visit), 48, and 72. Participants with post-entry HCV RNA ≥ lower limit of quantification (LLoQ) had deep sequencing of NS5A and NS5B genes performed. Phylogenetic analysis distinguished between reinfection and treatment failure. Reinfection rates per 100 person-years (PYS) were calculated with 95% confidence interval (CI) constructed using Poisson distribution. RESULTS:Of 397 participants with post-entry HCV RNA, 29 had ≥LLoQ and available sequencing data. Of those 29, 5 participants initially designated as non-SVR, and 12 participants initially attaining SVR (evaluated at week 24) were determined to have reinfections (total 17 reinfections) (reinfection rate 3.9/100 PYS [95% CI, 2.4-6.2]). All 17 participants with HCV reinfection were male (13 MSM and 15 with HIV). Of 29 had ≥LLoQ, 12 were identified as treatment failure. SVR (excluding reinfections) in presence and absence of baseline RAS was 93.5% (43/46) and 97% (337/346), respectively, with an overall SVR rate of 97.0% [95% CI, 94.8-98.3] (385/397). CONCLUSIONS:Accounting for reinfections, SVR in MINMON was 97.0% further supporting simplified HCV treatment. No significant difference in SVR was found by baseline velpatasvir RAS. The high reinfection rate, especially among men who have sex with men (MSM) with human immunodeficiency virus (HIV), underscores the need to scale-up evidence-based interventions to reduce reinfection. CLINICAL TRIALS REGISTRATION:NCT03512210.
