Advances in the HIV Care Cascade
Updated Data on the HIV Care Cascade and Modeling Future Trajectories
Beyrer delivered a plenary session (Abstract 17) on the global HIV/AIDS pandemic and presented new data modeling the impact of discontinuation of the President's Emergency Plan for AIDS Relief (PEPFAR). Mishra and colleagues developed a model to predict the impact of disruptions in PEPFAR services on HIV incidence in sex workers in Eswatini over the coming decade. In their worst-case scenario model, which aligns with current conditions of abrupt PEPFAR termination, disruptions of HIV prevention and treatment would lead to a rise in HIV incidence from below 0.5 per 1000 person-years to more than 1.5 per 1000 person-years. Bickerman and colleagues modeled the relative increase in new HIV infections related to the removal of PEPFAR-funded preexposure prophylaxis (PrEP) for key populations in sub-Saharan Africa. Countries with higher coverage of PEPFAR-funded PrEP services could see an increase in new infections by 30% among female sex workers, 20% among men who have sex with men (MSM) and transgender women, and 15% among people who inject drugs (PWID). Beyrer noted that, because the US Agency for International Development (USAID) serves as the implementing agency for approximately two-thirds of PEPFAR funding, we should not expect PEPFAR to be effective in the absence of USAID support.
Further modeling data to support concerns with the abrupt discontinuation of PEPFAR were highlighted by Imai-Eaton (Abstract 24) with data from the Thembisa model of the South African HIV epidemic. This model shows HIV incidence declining throughout the next 80 years if current prevention and treatment levels are sustained. However, if primary prevention efforts are discontinued and condom use declines, the model predicts HIV incidence to stabilize between 2 and 4 new cases per 1000 population between the years 2080 and 2100. The model further predicts that if these reductions in HIV prevention are combined with reduced HIV testing and access to antiretroviral therapy (ART), incidence in South Africa could increase to 10 per 1000 population by the year 2100, which is well above current levels.
Success in ending the HIV epidemic requires effective monitoring of the UNAIDS 95-95-95 targets. Moolla and colleagues used data from 7 sites participating in the South African IeDEA (International epidemiology Databases to Evaluate AIDS) cohorts to assess virologic suppression, defined as HIV-1 RNA levels less than 1000 copies/mL, among adults starting ART between 2005 and 2023 (Abstract 129). The analyzed cohort included 386,053 participants with 2.18 million years of followup data, a median of 5.1 years per person. Median age at ART initiation was 34.8 years (interquartile range [IQR], 28.8-42.2), 65.1% were women, and 7.5% initiated a dolutegravir (DTG)-based regimen. They found the trend in viral suppression steadily improved from a low of 86.4% in 2011, which was attributed to ART stockouts and increasing drug resistance, to the target of 95% viral suppression by 2022. DTG-based therapy was associated with decreased odds of virologic failure (adjusted odds ratio [aOR], 0.45; 95% CI, 0.42-0.50), regardless of whether the participants initiated ART on DTG or switched to a DTG-containing regimen. Prior treatment interruptions were associated with increased odds of virologic failure, even among those who return to care, whereas increasing age and female sex were protective. The COVID-19 pandemic did not lead to statistically significant increases in virologic failure. These data are limited by missing baseline CD4+ cell count and viral load information for a significant minority of the cohort, 51.4% and 38.5%, respectively, and also should be recognized as coming from a cohort of individuals receiving care rather than the general population. Considering the increased uptake of DTG-based regimens globally, these findings are encouraging as programs strive to reach the 95% viral suppression target.
Odiya and colleagues (Abstract 184) looked backward along the HIV care cascade by determining the care engagement status of individuals with viremia in a high HIV prevalence setting that has attained 90% virologic suppression. They used data from 34 communities participating in the Rakai Community Cohort Study in Uganda, including 17,841 individuals surveyed from 4 fishing villages with 40% HIV prevalence and 30 agrarian and semiurban communities with 12% HIV prevalence. An initial interview and HIV testing revealed that 3170 (17.8%) had HIV. Of those, 3134 had HIV viral load testing and 316 (10.1%) had viremia, defined as HIV-1 plasma RNA level higher than 1000 copies/mL. Respondents noted their care engagement status as: 170 (53.8%) newly diagnosed, 33 (10.4%) aware of status but never received ART, 105 (33.2%) engaged in care, and 8 (2.5%) disengaged from care. However, after clinical data extraction, laboratory testing for presence of ART, and a repeat interview with participants, 30.2% were reclassified. Notably, 49% of those engaged in care were reclassified as disengaged or never on ART, and 21% of new diagnoses were reclassified as previously diagnosed but never receiving ART. Misclassification was more common among men, those older than age 30 years, and those who reported being in care. This reexamination of care engagement status meant 82.9% of people with viremia were not in care, and 77.9% of those had never initiated ART. Investigators attempted to link all respondents not in care, with variable success. Of those newly diagnosed, 53.3% enrolled in care within 90 days, with a median time of 11.5 days (IQR, 0-36). Treatment initiation was less successful for those previously diagnosed but never initiating ART: 39.1% enrolled in care with median time to enrollment of 42.0 days (IQR, 10.5-123). Results of laboratory testing revealed that only 4.2% of those with viremia also had the presence of ART. These findings demonstrate that initial ascertainment of care engagement status among those with viremia may be influenced by social desirability bias among respondents and is likely an overestimate of care engagement.
Refining Care Cascade Metrics
Two studies explored how we measure steps along the care cascade from a different perspective. Yezdani and colleagues (Abstract 1081) used data from 13 NA-ACCORD (North American AIDS Cohort Collaboration on Research and Design) clinical cohorts between 2010 and 2021 to refine metrics for retention in care considering data linking fewer clinic visits with health. In this cohort of 95,428 adults with HIV contributing 546,805 person-years, they found that the adjusted cumulative incidence of death among those defined as retained in care by having at least 2 encounters 90 days apart within 12 months was higher than those who were not retained (adjusted hazard ratio [aHR], 1.17; 95% CI, 1.09-1.27). When the retention in care definition was expanded to include laboratory testing as an encounter, the association with incidence of death remained (aHR, 1.25; 95% CI, 1.16-1.36). These findings suggest that current metrics for care engagement in North America are predictive of adverse outcomes, rather than the positive step on the care cascade that has been assumed. Different methods for measuring retention in care may be needed to capture care engagement patterns that predict desired clinical outcomes.
The second study, from McManus and colleagues (Abstract 1078), examined differences between the Centers for Disease Control and Prevention (CDC) definition of viral suppression used by the Ryan White HIV/AIDS Program (RWHAP), last viral load measurement less than 200 copies/mL, and 3 more robust definitions of durable viral suppression. The analysis included a cohort of 618 adults aged 18 years or older who had established care at the University of Virginia RWHAP clinic at the beginning of 2018 and followed them up for 4 years (median age of the cohort, 51 years; 71% male; 38% had an income below federal poverty level). Examining viral suppression metrics for each year of the study, they found that 94% to 95% of individuals met the CDC definition of viral suppression. However, when they applied 3 alternative metrics of durable viral suppression, all viral loads under 200 copies/mL with: (1) 1 viral load required; (2) 2 viral loads required; or (3) 2 viral loads more than 90 days apart required, annual rates of viral suppression dropped. The least stringent metric of durable viral suppression, all viral loads under 200 copies/mL with at least 1 measurement during the year analyzed, showed 87% with durable viral suppression in 2018, 89% in 2019, 90% in 2020, and 92% in 2021. When 2 viral loads more than 90 days apart were required, the prevalence of durable viral suppression annually ranged from 83% in 2020 to 91% in 2021. They also conducted a sensitivity analysis that imputed detectable viral loads for those cohort members who were missing the required number of viral loads for each metric, with the anticipated effect of further reducing estimates of durable viral suppression. These findings demonstrate that national metrics based on the most recent viral load are likely overestimating the prevalence of viral suppression over time, and more robust metrics for viral suppression could provide further insight into this stage of the care cascade.
Specific Populations and the HIV Care Cascade
Mobility, which often leads to transitions between clinics, is recognized as a barrier to HIV care continuity. Rosen and colleagues (Abstract 186), who previously established that transitions between clinics can undermine HIV care engagement using proxy measures for clinic switch,1 used data from the PHIA (Population-Based HIV Impact Assessment) Project to examine direct measures of clinic switching and its impact on viral load across 7 countries in sub-Saharan Africa. Data collected in household surveys between 2020 and 2023 on 14,017 individuals from 11,642 households revealed that the pooled prevalence of clinic switching was 14% in the past year, with a range of 9.8% to 21.4%. Predictors of clinic switching included: age younger than 40 years (aOR, 1.28; 95% CI, 1.07-1.54); mobility in the past year (aOR, 1.47; 95% CI, 1.18,-1.84); more than 12 months on ART (aOR, 1.62; 95% CI, 1.22-2.22); and travel-related barriers to ART care (aOR, 1.23; 95% CI, 1.02-1.48). Multimonth dispensing of ART was inversely associated with clinic switching for 3- to 5-month prescriptions and for 6-month and longer prescriptions, compared with prescriptions shorter than 3 months. Overall, clinic switching was not statistically significantly associated with any of the 3 virologic failure outcomes: HIV-1 plasma RNA levels higher than 50, 200, or 1000 copies/mL, but was associated with missing 2 or more doses of ART in the past month by self-report (aOR, 1.29; 95% CI, 1.031.61). In a stratified analysis, women were more likely to experience viral load higher than 50 copies/mL and ART nonadherence. Living in a rural setting and age younger than 40 years were also associated with ART nonadherence. This analysis is limited by the cross-sectional study design and temporal associations between clinic switch and virologic outcome cannot be described. Despite this, these data underscore the need for processes to support planned and unplanned clinic switches, as they may be associated with poor HIV care outcomes.
Hemery and colleagues conducted a longitudinal analysis of time to ART initiation among 815 transgender women with HIV (WWH) within the French Hospital Database on HIV (Abstract 1110). They used Cox multivariable models to identify predictors of time to ART initiation after care entry for 3 periods: 1997 to 2004, 2005 to 2012, and 2013 to 2022. Most women (65.7%) were originally from Latin America, 70% were in Paris, and more than 99% listed sexual intercourse with a cisgender man as their mode of HIV acquisition. There was a statistically significant decrease in the proportion entering care with CD4+ counts below 200 cells/µL from 30.9% in the 1997 to 2004 period to 22% in the 2013 to 2022 period (P = .0005). There were also statistically significant decreases across the 3 periods in the number of months between: HIV diagnosis and care system entry; care system entry and ART initiation; and ART initiation and viral suppression below 200 copies/mL (P < .001 for all). Factors associated with more rapid ART initiation included AIDS status in the 2 earlier periods but not between 2013 and 2022. In the most recent period, those born outside of France took longer to initiate ART after care system enrollment than those born in France. These data demonstrate encouraging improvements over time in months to initiate ART for transgender WWH in France, but highlight continued challenges, particularly for those not born in France.
Leveraging Machine Learning and Artificial Intelligence to Predict HIV Care Cascade Outcomes
Marcus led an outstanding session (Themed Discussion 09) on leveraging machine learning for insights into care. Martin and colleagues used machine learning to predict people at risk for falling out of care, which continues to pose a challenge to the HIV care continuum in the US (Abstract 1307). They highlight the limitations of current data to care efforts, which follow up after a lapse of 12 months when many individuals have moved, transferred care to a clinic that does not report data into the system, been incarcerated, or died. The investigators constructed a Random Forest model using public health data from San Diego County, California, including longitudinal viral load and CD4+ cell count data from 2017 to 2022, hypothesizing that past viral load data may be predictive of future lapses in care. The analysis cohort was predominantly men (90%), White (45%), and between 40 and 60 years of age (48%). They examined performance of the model for prediction of the primary outcome, unsuppressed viral load in the 12 months following an index viral load, on training, refinement, and validation datasets. The validation dataset reached a low sensitivity for prediction of unsuppressed viral load of 33.8%, but a high specificity of 96.7%. The model found viral load testing variables, such as the proportion of suppressed or unsuppressed viral loads over time, were more strongly associated with the outcome than standard demographic variables like age, HIV exposure category, or race. The investigators believe that incorporating historic viral load and CD4+ cell count data could improve the ability of current data-to-care programs in predicting those at risk for negative outcomes.
Mirugwe and colleagues compared the effectiveness of various machine learning models including Decision Tree, Random Forest, AdaBoost, and XGBoost, with a deep learning, transformer-based model, Bidirectional Encoder Representations from Transformers (BERT), in identifying people with HIV (PWH) at highest risk for missing HIV treatment appointments, using data from Uganda's electronic health record (EHR) (Abstract 1309). The retrospective analysis included 66,206 PWH (65.1% women; median age, 41 years) who participated in 1.48 million clinical visits. The BERT model predicted missed appointments with 94.8% accuracy and 97.1% precision, with an area under the curve (AUC) score of 0.96 and an F1 score of 94.2%, performance parameters superior to the other 4 models examined. Similar to the model described in Abstract 1307, longitudinal treatment history emerged as a stronger predictor of future care disengagement than age or sex.
Focused on a similar goal, Wei and colleagues enhanced a pre-trained large language model, LLaMA 3.1, using data from January 2018 to June 2023 in Tanzania's National HIV Care and Treatment program's EHR (Abstract 1313). They extracted data from 4,809,765 records for 261,192 individuals in care and compared the enhanced large language model with 2 other machine learning models: a supervised machine learning model and a zero-shot learning model without any HIV specific information. They recruited HIV specialists from Tanzania to assess the accuracy of the enhanced large language model's reasoning and predictions. After internal and external validation, the enhanced large language model could identify individuals in the highest quartile of risk for 3 outcomes of interest: ART nonadherence, non-suppressed viral load, and loss to follow up (LTFU) with 73% accuracy, compared with less than 25% accuracy for the other 2 models. The investigators emphasized the ability for large language models to aid in the prediction of those at high risk for adverse care outcomes and to guide targeted interventions.
Estimating Morbidity and Mortality Among People With HIV
One important metric of progress in addressing the HIV epidemic is reductions in overall and HIV-associated mortality among PWH. Ambia and colleagues (Abstract 128) examined cause-specific mortality trends in northwestern Tanzania between 1995 and 2016, using an existing health and demographic surveillance system and verbal autopsy reports to assess cause of death. Their analysis included data from 68,878 individuals followed up for more than 276,265 person-years in a setting of 5.4% HIV prevalence. In this population, there were 2074 deaths during the observation period: 18.5% in PWH, 30.3% in people without HIV, and 51.2% in people of unknown HIV status. They found the overall mortality rate for PWH was 10-times that of those without HIV. The investigators were able to ascertain cause of death for 69% of all deaths. In a regression model adjusting for age and sex, the investigators demonstrated that all-cause mortality decreased significantly during 3 phases: the pre-ART years (1995-2004); the years during ART introduction (2005-2008; aHR, 0.7); and in the years in which ART was widely available (2009-2016; aHR, 0.25). Mortality rates from HIV, AIDS, and tuberculosis (TB) declined among PWH in each phase, but these remained the leading causes of death among PWH, even in the phase of widely available ART. All-cause mortality rates in people without HIV remained stable over time, implying that ART rollout improved mortality for PWH in this region.
Data from the US gathered in the HOPS (HIV Outpatient Study) between 2007 and 2022 also demonstrated low mortality rates but continued impact of advanced HIV disease (Abstract 1161). This retrospective cohort analysis included 4859 participants who had received at least 2 visits and ART for at least 6 months during the observation period (median participant age, 43 years [IQR, 33-49]; 20% women). The mortality rate was stable over time: 1.3 per 100 person-years in 2007 (95% CI, 0.9-1.9) and 1.1 per 100 person-years in 2022 (95% CI, 0.7-1.8), and most deaths were from non-HIV-related causes. Predictors of overall mortality included having public insurance, history of injection drug use, and being of non-Hispanic Black race. HIV-1 plasma RNA levels between 50 and 500 copies/mL and higher than 500 copies/mL were associated with increased mortality compared with levels less than 50 copies/mL. However, the strongest associations with mortality were CD4+ count under 200 cells/µL (adjusted relative risk [aRR], 7.10; 95% CI, 5.30-9.51) or between 200 and 350/µL (aRR, 2.74; 95% CI, 2.04-3.67). These data demonstrate the need to manage non-HIV-related risk factors for mortality alongside viral suppression to optimize health.
Mortality estimates based on EHR data can be biased by misclassification of patient outcomes, particularly by underestimating mortality in those who are LTFU. Van Heerden and colleagues (Abstract 130) addressed this challenge using data from TIER.net, South Africa's national HIV EHR system, combined with tiered and weighted sampling strategy to allow individual tracking of a subset of individuals LTFU in KwaZulu-Natal Province. They defined LTFU as any individuals who had not picked up medications for more than 3 months and, of 519 facilities, selected 36 clinics and 4482 individuals (19% of those LTFU) for tracing, including phone calls and neighborhood visits to ascertain vital status and link back to care. They succeeded in determining vital status for 70.4% of those selected for tracing, leading to an increase of “deceased” status within the cohort from 8.6% using only data from TIER.net to 13.2% with enhanced tracing. Weighted analysis estimated that 11% of those LTFU across the province were likely deceased. Systemic challenges in vital status ascertainment within the HER included duplications of records caused by weak linkages between clinics and untracked transfers. Those LTFU for longer periods were more likely to be deceased, as were men and those living in rural settings.
Interventions to Improve HIV Care Outcomes
Men in sub-Saharan Africa are more likely to be disengaged in care, and Dovel and colleagues (Abstract 185) tested 3 different interventions to support care engagement among men with HIV in Malawi in a non-blinded, randomized control trial. The interventions included: (1) a low-intensity intervention of male-sensitive counseling and navigation for ART initiation; (2) a high-intensity intervention with counseling plus home initiation of ART, and navigation for ART refills; and (3) a variable intensity stepped intervention that included counseling, and options for ongoing mentorship, home ART initiation, and navigation for ART refills. They included 569 men with HIV who were 15 years of age or older who had never initiated ART or were experiencing a treatment interruption (more than 90% of those enrolled). Their primary outcome, ART engagement at 9 months defined by starting ART within 90 days of engagement and not being more than 28 days out of care at 6 months, did not differ between the 3 groups. Engagement at 9 months was 76% in the low-intensity arm, 72% in the high-intensity arm, and 85% in the variable intensity/stepped intervention. The investigators noted that the high-intensity arm was more effective at reengaging participants in care and starting ART, but once participants were asked to transition from home-based ART back to the clinic, engagement declined. Qualitative interviews with a subset of participants revealed that they appreciated the human connection to the clinic received via counseling, which was available in all intervention arms, and the tailored approach. These data demonstrate that a low-intensity counseling intervention, focused on men and their specific needs, can lead to equivalent care engagement to higher-intensity interventions, although the lack of a standard-of-care arm does not allow comparison to the absence of any intervention.
Recent data from a cluster randomized trial in Tanzania showed that financial incentives given in the first 6 months after initiating ART were associated with improved retention in care and viral suppression at 6 months but did not have a statistically significant impact at 12 months.2 Based on a behavioral model that incentives given to those initiating ART, a group with high motivation for change, will lead to habit formation for routine behaviors such as care engagement, Winters and colleagues (Abstract 187) analyzed the impact of incentive timing within the same study. The trial randomly assigned 32 clinics and 1990 individuals initiating ART (mean days on ART at enrollment, 3.50; standard deviation [SD], 6.64) on a 1-to-1 basis to standard-of-care control or to 6 months of $10 US cash incentive monthly, conditional on appointment attendance. Primary outcomes were: missed appointments, defined as more than 4 days post a scheduled visit; low ART adherence, defined as under 90% of days covered by ART; and interruption in treatment for more than 28 days. Using Cox proportional hazard models adjusting for key individual characteristics and clinic-level clustering, they found a statistically significant difference in missed appointments between the 2 arms (aHR, 0.72; 95% CI, 0.59-0.88), but not in ART adherence (aHR, 0.80; 95% CI, 0.63-1.01) or interruption in treatment (aHR, 0.79; 95% CI, 0.60-1.05). They also noted large differences in care-engagement behavior at follow-up. The aHR for missing appointments was significantly lower in the first 6 months while financial incentives were in place (aHR, 0.51; 95% CI, 0.39-0.68), but immediately dropped after financial incentives were removed (aHR, 0.98; 95% CI, 0.75-1.29) in the last 6 months. The same patterns were repeated for the other 2 outcomes. These data are contrary to the investigators’ hypothesis that incentives could prompt habit formation for care engagement and suggest that incentives have little to no effect once they are removed.
Mody and colleagues pursued a different strategy to improve care engagement, designing a person-centered care (PCC) intervention for practitioners and clinic staff to improve patient experience. They presented further findings from a study that found an increase in retention in care after a PCC intervention in Zambia.3 This stepped wedge trial was conducted in 24 public health clinics and the intervention involved 3 components: training and in-clinic coaching for practitioners on PCC principles; systematic exit surveys for patients to measure clinic experience and feedback to staff to drive improvement efforts; and small performance-based incentives to the clinics. To better understand why the PCC intervention improved retention, the investigators looked at its effects on 2 outcomes: the incidence of return to care after an initial treatment interruption; and the incidence of repeat treatment interruption among those who returned to care. They included 120,701 clients (64,283 in the control arm and 56,418 in the intervention arm) with at least 1 clinic visit during the study period (August 2019 through November 2021) who then were more than 30 days late to a scheduled visit, meeting the definition for treatment interruption. Investigators found that the PCC intervention was associated with an increased incidence of return to care after treatment interruption (65.4% in the control arm and 71.1% in the intervention arm, with a statistically significant risk difference of 5.7%; 95% CI, 5.3-6.2). Among the 78,561 who returned to care, the PCC intervention was associated with decreased incidence of repeat treatment interruption from 55.3% in the intervention arm to 44.9% in the control arm and a statistically significant risk difference of -10.4% (95% CI, -11.1 to -9.8). They also included data on return to care after a repeat treatment interruption to explore the number of people in care overall at 1 year after 1 or more treatment interruptions, with a statistically significant difference from those in care in the control group (49.4%) and the intervention group (58.7%; risk difference [RD], 9.3%; 95% CI, 8.8-9.8). These findings demonstrate that focusing on clinician-client relationships can lead to measurable improvements in key HIV care outcomes in public health settings.
The Max Clinic in Seattle, Washington, is a low-barrier care clinic, a differentiated service-delivery model shown to improve rates of virologic suppression among PWH with numerous barriers to care.4 Dombrowski and colleagues (Abstract 1304) conducted a retrospective cohort analysis of 398 PWH enrolled in care, with at least 6 months of observation in the Max Clinic between December 2014 and December 2023. The cohort included 76% cisgender men, 21% cisgender women, and 4% transgender women. There was also a high burden of known risk factors for virologic failure (78% experienced unstable housing or homelessness, 73% substance use, and 28% severe mental illness). The overall likelihood of viral suppression increased from 35.6% before clinic enrollment to 60.3% post enrollment, with an unadjusted relative risk of 1.78 (95% CI, 1.61-1.96). Investigators explored interactions between predictors to develop generalized estimating equations to predict virologic suppression. They determined a 5-way interaction existed between sex, Max clinic enrollment status, and the 3 risk factors of housing insecurity, substance use, and severe mental illness, with probability of viral suppression increasing from 30% to 70% before and after Max clinic enrollment for individuals with all 3 risk factors, compared with 38% to 52% among those without risk factors. The investigators note that the benefits of the low-barrier clinic model are greatest among patients with the highest burden of risk for virologic failure.
Advances in Hepatitis B and C
Hepatitis C Epidemiology, Treatment, and Outcomes
Arens and colleagues (Abstract 108) analyzed hepatitis C trends among PWH from 1995 to 2023 among several sites across the US. Their primary outcome was incidence of new primary hepatitis C virus (HCV), defined as a newly positive hepatitis C antibody or plasma RNA, and they calculated incidence ratios (IRs) in 3 calendar periods. In their large, retrospective cohort of 20,917 PWH, 81% were cisgender men with a median age of 37 years. There were 42% who identified as non-Hispanic White and 41% who identified as non-Hispanic Black. There were 61% who were MSM and 6% were PWID. In this cohort, 54% were not virally suppressed, with an HIV viral load above 200 copies/mL. They identified 1276 new cases of primary hepatitis C (6.1%), with an incidence ratio (IR) of 7.1 (95% CI, 6.7-7.5) per 1000 person-years. When looking at IRs in MSM across the 3 periods, the IR increased by more than 2-fold from 1995 to 2005 to 2015 to 2023. There was no increase in IRs in PWID across the calendar periods. The authors concluded that HCV incidence in MSM needed to be studied further to understand the increase in hepatitis C incidence in this particular population.
Arshad and colleagues (Abstract 640) presented pharmacokinetic data for a solid long-acting injectable for hepatitis C with solid glecaprevir and pibrentasvir. They injected solid pellets containing 27 mg of glecaprevir and 27 mg of pibrentasvir into rats and monitored plasma concentration of the drugs over 13 weeks. They observed rapid maximum plasma concentrations 2 hours post administration for glecaprevir and 11 hours post administration for pibrentasvir. They observed detectable plasma concentrations of the antivirals during the full 13 weeks of the study. They found similar pharmacokinetics between the solid formulation and the previously tested liquid injectable glecaprevir and pibrentasvir presented at Conference on Retroviruses and Opportunistic Infections (CROI) 2024.5 They did not observe adverse effects in the rats, including no observed behavioral issues, weight loss, or injection-site abnormalities. This solid formulation of glecaprevir/pibrentasvir represents a possible powerful long-acting tool in the treatment and eradication of hepatitis C.
Workowski and colleagues (Abstract 748) looked at efficacy and safety of an 8-week course of glecaprevir/pibrentasvir in PWH and acute hepatitis C in a phase IIIb, multicenter trial. The primary efficacy endpoint was sustained virologic response 12 weeks after treatment course (SVR12). They enrolled PWH on ART with a CD4+ count of 00 cells/µL or higher with a new diagnosis of hepatitis C. A total of 87.3% of the participants had viral suppression at baseline; 97.2% (no CIs given) of PWH and acute hepatitis C achieved SVR12 in the intention-to-treat (ITT) arm, and 100% (no CIs given) of participants achieved SVR12 in the modified ITT arm. There were no observed virologic failures, and there were no adverse events related to the glecaprevir/pibrentasvir and no deaths. There was also no effect on HIV viral suppression while on glecaprevir/pibrentasvir. The authors concluded that an 8-week regimen of glecaprevir/pibrentasvir is safe and effective for acute hepatitis C in PWH currently on ART.
Lang and colleagues (Abstract 749) examined factors that contributed to (lack of) treatment initiation in PWH and hepatitis C by looking at NA-ACCORD cohort data from 2014 to 2021. The cohort consisted of 6300 PWH with HCV viremia throughout a study period of 14,050 person-years. Only 58% of participants started direct-acting antivirals (DAAs). Reassuringly, the gap between those eligible for DAA therapy and those starting DAA treatment decreased from 77% at year 1 to 6% at year 8 of follow-up. Factors associated with decreased update of DAAs included injection drug use (aHR, 0.83; 95% CI, 0.74-0.93), being non-Hispanic Black (aHR, 0.82; 95% CI, 0.72-0.89), being Hispanic (aHR, 0.86; 95% CI, 0.75-0.98), smoking (aHR, 0.76; 95% CI, 0.69-0.84), and detectable viremia (aHR, 0.60; 95% CI, 0.54-0.66). A higher fibrosis-4 (FIB-4) score was associated with increased rates of DAA use (aHR, 1.22; 95% CI, 1.10-1.36). The study authors concluded that targeted approaches to reach these vulnerable groups are needed to achieve the goal of eliminating hepatitis C.
Solomon and colleagues (Abstract 750) hypothesized that HCV reinfection may be masking true SVR12 rates in high-risk populations, such as PWID. They observed HCV reinfection in treated PWID in India. The median age of their cohort of 1747 individuals was 29 years, and 57% were PWID. They observed 715 new HCV reinfections, with an average reinfection rate of 44.3 per 100 person-years. The reinfection rate was highest right after achieving SVR12 and tended to decrease during the study period. Variables associated with a high risk of HCV reinfection included increased injection use, sharing needles, being coinfected with HIV (with or without ART), and younger age, indicating high-risk groups who could benefit from additional services to prevent HCV reinfection. McFall presented a study (Abstract 751) from the same study group showing high rates of hepatitis C infection with unfortunately no change in aviremia in their cohort of PWID in India from 2012 to 2024, suggesting more programmatic resources are needed to meet 2030 targets to eliminate HCV.
Liegeon and colleagues (Abstract 752) examined the effects of an HCV test-and-treat intervention in their ANRS-PREVENIR (Prevention of HIV in “Île-de-France”) cohort on hepatitis C incidence in MSM throughout separate study periods. They looked at incidence before the implementation from May 2017 to January 2020, during implementation of the test-and-treat strategy from January 2020 to June 2021, and then post implementation from June 2021 to May 2024. The cohort numbered 3050 in the preimplementation period, 2300 in the implementation period, and 1827 in the postimplementation period. The hepatitis C incidence was 7.9 per 1000 person-years (95% CI, 5.6-11.1) in the preimplementation period, which dropped to 5.1 per 1000 person-years (95% CI, 3.1-8.4) during the implementation period, and further decreased to 2.9 per 1000 person-years (95% CI, 1.6-5.3) in the postimplementation period (P < .01), suggesting that a test-and-treat campaign is effective in decreasing hepatitis C incidence.
Estadt and colleagues (Abstract 754) set out to understand the prevalence of comorbidities in Medicaid participants with chronic hepatitis C from 2016 to 2020 within 12 months of diagnosis (or index date). A total of 59% of the Medicaid patients with hepatitis C were between 45 and 64 years of age. Based on their models, younger individuals (aged 18-44 years) were less likely to be diagnosed with HIV (adjusted prevalence ratio, [aPR], 0.56; 95% CI, 0.43-0.73) or alcohol use disorder (aPR, 0.68; 95% Cl, 0.58-0.79), but more likely to be diagnosed with chlamydia (aPR, 5.52; 95% Cl, 4.23-7.21), substance use disorder (aPR, 1.57; 95% Cl, 1.50-1.65), have at least 1 drug overdose (aPR, 1.63; 95% Cl, 1.53-1.75), and have at least 1 mental health disorder (aPR, 1.17; 95% Cl, 1.14-1.21). They observed high rates of substance use disorders (53.6%) and mental health disorders (49.7%) in the overall cohort. The authors suggest that targeted services addressing some of these comorbidities would be of benefit when linking patients to hepatitis C treatment.
Corma-Gomez and colleagues (Abstract 798) developed and validated a model for predicting hepatocellular carcinoma (HCC) in patients with treated HCV. Their cohort included more than 1000 patients in Spain with hepatitis C who had been treated with DAAs, had achieved an SVR, and had a liver stiffness test result of 9.5 kPa or higher. Their cohort included nearly 600 PWH (57%), with a median age of 52 to 55 years, and an overall diagnosis rate of 3% for HCC in their cohort during the study period. Their machine-based learning model included age, liver stiffness, Fib-4 score, gamma-glutamyl transferase (GGT) and albumin levels at time of SVR, as well as fasting glucose, and GGT levels and liver stiffness at 12 months post SVR to classify individuals as at low HCC risk vs not at low risk. When they tested this in their “model discovery” cohort (which comprised 67% of the total cohort), the sensitivity of the model was 100% (95% Cl, 85-100) with a negative predictive value for HCC of 100% (95% Cl, 99-100), and a Harrell's C index of 0.834. In their validation cohort (33% of the total cohort), their model's sensitivity was 91% (95% Cl, 59-100) with a 99% negative predictive value (95% Cl, 97-100) and a Harrell's C index of 0.857. In the validation group, 1 person in the low-risk group developed HCC; incidence was 0.08 per person-years, which fell below the 1% threshold where screening for HCC is no longer cost-effective. The study authors conclude that their machine-based learning model can effectively predict who is at low risk of HCC after treatment for hepatitis C and SVR, and that the incidence of HCC in this group does not support screening due to lack of cost-effectiveness.
Hepatitis B Epidemiology, Treatment, and Outcomes
Thio and colleagues (Abstract 109) looked at hepatitis B surface antibody (HBsAb) response in PWH and those without HIV who had incident hepatitis B infection. They examined HBsAb titers and neutralization capacity. Only 4 PWH were on antihepatitis B medications at the time of the incident infection. Following HBsAb titers longitudinally, they found that HBsAb titers were significantly higher at 6 and 30 months after infection in persons without HIV than in PWH (P <.001). In fact, in those without HIV, the HBsAb titers rose during the study period, whereas in those without HIV, the titers waned (P <.0001). This trend was also seen in the neutralizing capacity of the HBsAb. Interestingly, when including unsuppressed viral load (HIV viral load >10,000 copies/mL) or CD4+ count less than 350 cells/µL in the model, neither was associated with a statistically significant difference in antibody response or neutralizing capacity of HBsAb in PWH.
Marks and colleagues (Abstract 112) presented seroprotective response (SPR) durability results from the ACTG (AIDS Clinical Trial Group) A5379 BEeHIVe (B-Enhancement of HBV Vaccination in Persons With HIV) trial comparing Heplisav-B adjuvanted HBV vaccine with CpG (HepB-CpG) with standard hepatitis B vaccine in PWH. Their primary analysis previously showed that 2 doses and 3 doses of Hep-CpG had a superior response to standard hepatitis B vaccine in group A (prior nonresponders)6 as well as excellent SPR in group B (vaccine-naive PWH).7 Seroprotective durability (defined as HBsAb titer >10 mIU/mL) was determined at the end of the study at week 72, which was 48 weeks after the third dose of Hep-CpG if receiving 3 doses, and 68 weeks after the second dose of Hep-CpG if receiving 2 doses. In group A, which included 561 individuals in 10 countries, the SPR was highest in the group that received 3 doses of Hep-CpG at 97.2%, compared with 86.1% in the 2-dose Hep-CpG group. The SPR was lowest in the standard hepatitis B vaccine 3-dose series, at 57.5%. In examining the SPR, there was a 39.2% difference (95% Cl, 30.0-47.9) between the 3-dose HepB-CpG vs the standard 3-dose series, a difference of 28.6% (95% Cl, 18.0-38.4) between the 2-dose HepB-CpG vs standard hepatitis vaccine 3-dose series, and 11.0% (95% Cl, 4.3-18.1) between 2 and 3 doses of HepB-CpG. Primary vaccine response was predictive of SPR; among those with HBsAb titer more than 1000 mIU/mL at the primary vaccine response time, all participants had SPR at week 72. In group B, 97.3% of participants had SPR at the end of the study (95% CI, 90.7-99.3). Ultimately, the authors concluded that higher rates of SPR occurred with HepB-CpG vs standard hepatitis B vaccine in group A, and higher rates of SPR with 3 doses than with 2 doses of HepB-CpG, supporting the use of 2- or 3-dose HepB-CPG in PWH.
In 2024, the World Health Organization (WHO) released updated hepatitis B treatment guidelines simplifying and expanding the number of people eligible for treatment. Wembulua and colleagues (Abstract 767) compared eligibility for hepatitis B treatment under the prior 2015 WHO guidelines and the updated 2024 guidelines in their SEN-B cohort, a cohort of persons with hepatitis B in Senegal. They included 748 treatment-naive individuals (median age, 31 years; 53% men; 98% hepatitis B e antigen [HBeAg] negative). Twenty-five percent of women and 30% of men had HBV DNA levels higher than 2000 IU/mL and 2.3% of women and 6.6% of men had an aspartate transaminase (AST) to platelet index ratio (APRI) score above 0.5. Overall, 3.2% of women and 18.2% of men had liver stiffness measurements higher than 7 kPa. Overall, based on liver stiffness, 6.3% of their cohort was eligible for DAAs under the WHO 2015 guidelines and this number increased to 34.8% under the 2024 guidelines (P <.001). Based on the APRI score, 5.1% of their cohort was eligible under 2015 guidelines and 29.8% were eligible under 2024 guidelines (P < .001). They calculated 9.7% of their cohort was previously ineligible for treatment but became eligible in the first year of clinical follow-up. This represents an important cohort to reevaluate eligibility for hepatitis B treatment according to new guidelines.
Loeb and colleagues (Abstract 768) compared hepatitis B surface antigen (HBsAg) expression in PWH with people with HBV monoinfection on DAA treatment. They examined HBsAg using a highly sensitive assay that is 10-times more sensitive than commercial assays, along with HBV DNA and HBsAb levels in PWH enrolled as part of a trial. The median age of their 257 participants was 30 years, 28% were on ART, median CD4+ count was 379 cells/µL, and only 16% were virally suppressed with an HIV viral load under 150 copies/mL. A total of 37% of participants had a positive highly sensitive HBsAg assay (value, 1-10) that would have been negative by commercial assays along with detectable HBV, consistent with occult HBV. In this group, median anti-hepatitis B core antibody (HBcAb) level tended to be higher than those with a positive highly sensitive HBsAg assay and undetectable HBV level (aOR, 1.57; 95% CI, 1.26-1.95; P < .05) and less likely to be on ART (aOR, 0.1; 95% CI, 0.02-0.64; P <.05). Interestingly, neither viral suppression nor CD4+ cell count was associated with occult hepatitis B. The authors conclude that these data support the hypothesis that HBsAg can exist as a continuum (instead of positive or negative), especially in PWH.
Hepatitis B reactivation in PWH with isolated HBcAb positivity on non-hepatitis B active ART is a common concern as more and more people switch to tenofovir (TFV)-sparing therapy, including long-acting injectable regimens. A series of abstracts used cohort data to examine this question. Arcos-Rueda and colleagues (Abstract 770) looked at 77 PWH with isolated HBcAb confirmed to be on TFV-sparing regimens in Spain. They observed no hepatitis B reactivation after a median time of 2.1 years, including 20 participants on long-acting ART or regimens without TFV or lamivudine (3TC). It was reassuring to find that the risk of hepatitis B in this patient population was quite low. Ring and colleagues (Abstract 771) performed a systematic literature review and identified 4 cohort studies and 7 case reports looking at HBV reactivation in PWH who were HBsAg negative and HBcAb positive who switched to completely hepatitis B-sparing regimens. Among the 4 cohorts with 6093 individuals, there were 99 hepatitis B reactivations, at a rate of 1.6% (95% CI, 1.3-2.0). The time to reactivate varied widely, from 3 weeks to 14 months. The risk of reactivation tended to be higher in patients who had no or low HBsAb titers (below 10 IU/L) with a range of 3.2% to 14%, than in those who had HBsAb titers of 10 IU/L or higher, with a range of 3.2% to 14% (no P-values given). The hepatitis B flares resolved with restart of hepatitis B-active agents. Mohareb and colleagues (Abstract 772) shared findings of hepatitis B outcomes among a French cohort with a median 14.5 years of follow-up, divided into 3 groups: those who were on TFV, those who discontinued TFV, and those who never started on TFV. Baseline characteristics such as age, median duration of ART, and CD4+ cell count were similar among the 3 groups. A detectable HBV viral load above 2000 IU/mL in PWH not on TFV was associated with detectable HIV RNA (P = .041), lower CD4+ cell count (P = .027), positive HBeAg (P = .004), and positive anti-hepatitis D antibody (P = .001). HBsAg loss was observed in 23% and 44% of individuals who discontinued TFV or never initiated TFV, respectively. These risk factors associated with hepatis B viremia may help risk stratify patients with coinfection who are at higher risk of HBV reactivation off TFV therapy.
Kasone (Abstract 776) presented performance characteristics of the point-of-care Xpert hepatitis B test in HBsAg-positive pregnant women in Uganda. In a cohort of 123 pregnant women of at least 24 weeks gestational age in Uganda, they observed a 100% agreement with the existing reference (laboratory) testing, with 100% sensitivity, 100% specificity, and an r coefficient of 0.99. The median turnaround time from collecting the sample to the result release was 1 hour. Sixty-one percent of mothers received results the same day and 20% of mothers were identified to have a hepatitis B viral load greater than 200,000 IU/mL and were initiated on same-day TFV, indicating feasibility for same-day testing and treatment.
Bertoni and colleagues (Abstract 779) evaluated and validated a point-of-care ultrasound model for finding cirrhosis and HCC in patients with hepatitis B in Zambia. They enrolled 202 patients in the study, 28% of whom were coinfected with HIV and HBV. Thirty percent had significant fibrosis by transient elastography and 12% had cirrhosis; 4% overall were found to have liver lesions. The model did not perform well for detecting significant fibrosis, with a sensitivity of 28.3% (95% CI, 17.5-41.5), specificity of 99.3% (95% CI, 96.1-100), and receiver operating characteristic (ROC)-AUC of 0.64. The model performed better for cirrhosis, with sensitivity of 56% (95% CI, 34.9-75.6), specificity of 97.7% (95% CI, 94.3-99.4), and ROC-AUC of 0.77. The model was better for detecting liver lesions, with a sensitivity of 85.7% (95% CI, 42.1-99.6), specificity of 93.5% (95% CI, 88.996.6), and ROC-AUC of 0.90. The authors concluded that point-of-care ultrasound could be a useful tool to detect cirrhosis and liver lesions in individuals with hepatitis B, especially in low- and middle-income countries (LMICs).
Advances in Antiretroviral Therapy
Clinical Trials of Novel Antiretroviral Agents
VH-184. Rogg and colleagues presented data on VH-184, a third-generation integrase strand transfer inhibitor (InSTI) that is active against isolates that are resistant to currently available InSTIs (Abstract 152). They enrolled 23 participants who were naive to ART and were randomly assigned to receive VH-184 at 10 mg (n = 6), 50 mg (n = 7), 300 mg (n = 7), or placebo (n = 3) on days 1, 4, and 7 (median age, 32 years; 86% men; 68% White). The mean baseline HIV-1 RNA ranged from 4.3 to 5.0 log10 copies/mL. The primary endpoint was maximal change in plasma HIV-1 RNA level through day 10, after which participants received locally available ART. The 50 mg and 300 mg groups experienced a maximal decrease of 2.2 and 2.3 log10 copies/mL, respectively. The 10 mg group experienced a 1.2 log10 decrease and the placebo group experienced a 0.3 log10 decrease. The authors noted that the decrease in the higher-dose groups was similar to results from studies of other InSTIs. No emergent resistance was observed, and no adverse events led to premature treatment discontinuation. Of note, VH-184 is being developed as a long-acting injectable product.
VH-499. Benn and colleagues presented data on VH-499, a new HIV-1 capsid inhibitor being developed as long-acting antiretroviral agent (Abstract 153). They enrolled 23 participants who were naive to ART and were randomly assigned to receive VH-184 at 25 mg (n = 7), 100 mg (n = 6), 250 mg (n = 7), or placebo (n = 3) on days 1 and 6 (median age, 31 years; 83% men; 65% White). The mean baseline HIV-1 RNA level ranged from 4.2 to 5.2 log10 copies/mL in the 4 arms. The primary endpoint was maximal change in plasma HIV-1 RNA level through day 11, after which participants received locally available ART. The 25 mg, 100 mg, and 250 mg groups experienced a maximal decrease of 1.8. 1.8, and 2.2 log10 copies/mL decline, respectively, and the placebo group experienced a 0.2 log10 decline. One person in the 25 mg group had an emergent capsid inhibitor resistance-associated mutation. No adverse events resulted in premature treatment discontinuation.
N6LS and cabotegravir. Taiwo and colleagues presented data on N6LS, a broadly neutralizing CD4-binding site antibody, given with long-acting cabotegravir as maintenance ART in adults who were stably suppressed on an oral regimen (Abstract 203). Participants were required to have sensitivity to N6LS on a phenotypic assay. One hundred twenty-five participants were randomly assigned 2:2:1 to N6LS intravenously plus monthly intramuscular cabotegravir, N6LS, and hyaluronidase subcutaneously plus monthly intramuscular cabotegravir, or to continue oral ART. Participants had a median age of 53 years, 83% were men, and 63% were White. The primary endpoint was plasma HIV-1 RNA level of 50 copies/mL or higher according to the US Food and Drug Administration (FDA) snapshot regimen. This occurred in 2 (4%) of the intravenous N6LS group and 3 (6%) of the subcutaneous N6LS group compared with none in the oral ART group. Two participants in each N6LS group met protocol-defined confirmed virologic failure. The group that had subcutaneous injections had more injection-site reactions, including some that led to withdrawal of assigned therapy. The authors noted that the intravenous formulation was chosen for further evaluation in part of this ongoing clinical trial.
Lenacapavir and 2 broadly neutralizing antibodies. Ogbuagu and colleagues presented data on every-6-month lenacapavir (LEN) with 2 broadly neutralizing antibodies (bNAbs), teropavimab (TAB) and zinlirvimab (zab), as maintenance ART in adult PWH with viral suppression on oral ART (Abstract 151). Eighty participants were randomly assigned 2:1 to change to their ART to LEN/TAB/ZAB or to maintain current oral ART. Participants were required to be susceptible to TAB and ZAB and 50% of potential participants were excluded due to nonsusceptibility. Participants were 85% male, 55% White, and 25% Hispanic, and the median CD4+ count was above 700 cells/µL. The primary endpoint was HIV-1 RNA below 50 copies/mL according to the FDA snapshot algorithm at week 26. This was achieved by 96% of participants in both arms. One participant in the LEN/TAB/ZAB arm experienced virologic failure occurring at week 24, with emergent resistance to LEN and loss of ZAB susceptibility. The pharmacokinetic analyses showed that concentrations were in the target ranges throughout the dosing for the LEN/TAB/ZAB regimen. Antidrug antibodies to TAB and ZAB were observed in 11% and 17%, respectively, but this did not impact antibody pharmacokinetics, safety, or efficacy. There were no grade 3 treatment-related adverse events in either arm and no treatment-related adverse events leading to discontinuation. Injection-site reactions to LEN were common. Longer-term results are forthcoming. The authors concluded that these data are supportive of further investigation of this twice-yearly ART regimen.
Dual antibodies as maintenance ART. Lalezari and colleagues presented data on the combination of 2 antibodies for the maintenance of viral suppression in adult PWH (Abstract 701). Participants received every-8-week infusions of TMB-365, an antibody targeting the host CD4+ receptor, thereby preventing HIV-1 entry, and TMB-380 (also known as VRC07-523LS), a bNAb targeting the CD4-binding region of HIV-1. Participants were not screened for susceptibility to the antibodies. Two participants exited the study early due to rash and were excluded. Eighteen participants completed the study and 17 had results available at week 24, at which time 16 (94%) had plasma HIV-1 RNA level below 50 copies/mL. Infusion-related adverse effects were common. The authors suggested that these data support further investigation of this combination for long-acting ART.
Doravirine/islatravir. There were 2 clinical trials of a fixed-dose combination of doravirine (DOR) 100 mg and islatravir (ISL) 0.25 mg given daily as maintenance ART. Colson and colleagues (Abstract 204A) presented data from a clinical trial enrolling people who were stably suppressed on oral bictegravir/emtricitabine/TFV alafenamide (BIC/FTC/TAF) who were randomly assigned 1:2 to continue this regimen or switch to daily DOR/ISL. This trial enrolled 513 participants (median age, 47 years; 21% women; 61% White). Participants were required to be negative for chronic hepatitis B, although 26% had evidence of prior HBV infection by HBcAb positivity. The primary endpoint was plasma HIV-1 RNA level of 50 copies/mL or higher according to the FDA snapshot regimen at week 48. This occurred in 1.5% of the DOR/ISL group and 0.6% of the BIC/FTC/TAF group (difference, 0.9; 95% CI, -1.9 to 2.9). This met the study protocol's definition of noninferior virologic efficacy. Two participants in the DOR/ISL arm had protocol-defined confirmed virologic failure. No emergence of resistance was identified. The lymphocyte counts and CD4+ cell counts were similar between arms, confirming that the lower dose of ISL is not associated with lymphocyte depletion, and no safety concerns were identified. There were no differences in the mean weight change between groups. Among 86 participants with HBcAb positivity receiving DOR/ISL, 2 were identified to have low-level HBV viremia without antigenemia or elevated transaminase levels and 2 other participants developed acute HBV; there were no cases of low-level HBV viremia or acute HBV in the BIC/FTC/TAF group.
Orkin and colleagues (Abstract 204B) presented data from a similar clinical trial that enrolled PWH on a stable 2- or 3-drug oral ART regimen who were randomly assigned 1:2 to continue their oral ART regimen or switch to DOR/ISL. This trial enrolled 551 participants (median age, 51 years; 40% women; 39% White). Participants were required to be negative for chronic hepatitis B, although 29% had evidence of prior HBV infection by HBcAb positivity. The primary endpoint was plasma HIV-1 RNA level of 50 copies/mL or higher according to the FDA snapshot regimen at week 48. This occurred in 1.4% of the DOR/ISL group and 4.9% of the continued ART group (difference, -3.6; 95% Cl, -7.8 to -0.8). This met the study protocol's definition of noninferior virologic efficacy. The lymphocyte counts and CD4+ cell counts were similar between arms and no safety concerns were identified. There was more weight gain in the DOR/ISL group; this was driven by weight gain among those discontinuing efavirenz (EFV) or TFV disoproxil fumarate (TDF) (which suppress weight gain) switching to DOR/ISL. The authors for these studies concluded that the data provide evidence for the efficacy of this 2-drug, non-InSTI-containing regimen for those with stable virologic suppression.
SARS-CoV-2. Hayden and colleagues (Abstract 200) presented data from the SCORPIO-PEP (Stopping COVID-19 Progression With Early Protease Inhibitor Treatment - Postexposure Prophylaxis) trial, a randomized, placebo-controlled trial evaluating ensitrelvir for postexposure prophylaxis (PEP) for SARS-CoV-2 household contacts. Ensitrelvir is an oral SARS-CoV-2 3C-like PI, with a similar mechanism of action to the combination of nirmatrelvir and ritonavir (RTV) used for COVID-19 treatment in the US. The phase III trial enrolled 2389 household contacts, with the primary efficacy endpoint being the proportion of household contacts who became SARS-CoV-2 positive and symptomatic by day 10. The mean age in the modified ITT arm was 42.4 years and 37% had at least 1 high-risk factor for progression to severe COVID-19. In total, 71.1% were enrolled within 48 hours of symptom onset in the index patient. The proportion of patients in the ensitrelvir arm who developed symptomatic COVID-19 was significantly lower (2.9%) than the placebo group (9.0%), with a risk ratio (RR) of 0.33 (95% Cl, 0.22-0.49; P <.0001). In a subgroup analysis of patients at high risk, they observed a significantly decreased rate of symptomatic SARS-CoV-2 by day 10 in the arm (2.4%) compared with the placebo arm (9.9%), with an RR of 0.24 (P < .0001). They observed a similar rate of adverse events in the ensitrelvir and placebo arms, and there were no hospitalizations or deaths related to COVID-19. This was the first study showing efficacy of an antiviral for SARS-CoV-2 PEP.
Long-Acting Antiretroviral Therapy
CARES trial. Few data exist on long-acting cabotegravir (LA-CAB)/rilpivirine (RPV) in sub-Saharan Africa, where there is a larger diversity of HIV subtypes and preexisting nonnucleoside reverse transcriptase inhibitor (NNRTI) resistance mutations are more common. Kityo and colleagues (Abstract 202) presented 96-week results from CARES (the Randomized Trial of Cabotegravir and Rilpivirine Long-Acting in Africa). The investigators previously presented week-48 data that showed noninferiority of CAB-LA/RPV to continued oral ART.8 This trial enrolled those stably suppressed on a DTG- or NNRTI-containing regimen with no known history of virologic failure and without chronic hepatitis B. The study enrolled 512 participants (median age, 42 years; 58% women; 99.6% Black African). Testing of archived HIV-1 DNA showed that 3% harbored mutations conferring moderate- to high-level RPV resistance and 2% harbored mutations conferring moderate- to high-level CAB resistance. Participants were randomly assigned to continue oral ART or to start CAB-LA/RPV after an optional oral lead-in. Plasma HlV-1 RNA level was assessed every 24 weeks. Ninety-seven percent of the scheduled injections were given within the study-mandated windows. At week 96, 96.9% of the CAB-LA/RPV group and 97.3% of the oral ART group had plasma HlV-1 RNA levels of less than 50 copies/mL (difference, -0.4%; 95% CI, -3.1 to 2.0). This achieved protocol-specified noninferiority of virologic efficacy. The proportion with plasma HlV-1 RNA levels of 50 copies/mL or higher was similar between the 2 groups. Four participants receiving CAB-LA/RPV had virologic failure (3 had emergence of RPV and CAB resistance and 1 had unsuccessful resistance testing). The authors concluded that these results establish the longer-term efficacy and support the use of CAB-LA/RPV in sub-Saharan Africa.
Yearly lenacapavir. Singh and colleagues (Abstract 154) presented data on 2 new formulations of LEN to be dosed once yearly. Forty participants (65% male; 80% White; 90% Hispanic) received bilateral gluteal injections of the assigned formulation. The concentrations achieved were higher than twice-yearly subcutaneous LEN through 56 weeks for both formulations. The trough concentrations for the intramuscular formulations were more than 2-fold higher than those observed with twice-yearly subcutaneous LEN. Most participants reported no pain or mild pain, which typically resolved within a week. No nodules were observed. The authors noted that once-yearly LEN will be studied in a phase III PrEP trial.
Three-drug nanosuspension. Bender Ignacio and colleagues presented data on a novel nanoparticle suspension of 3 ART drugs: lopinavir (LPV), RTV, and TFV (Abstract 639). Twelve participants (8 men and 4 women) received a single injection and were followed closely. The LPV pharmacokinetics exhibited a prolonged halflife of 33 days, whereas the RTV and TFV concentrations were lower than expected. Of note, there was 1 instance of anaphylaxis and 1 adverse event of lip swelling. The authors noted that these events were not immunoglobulin E (IgE) mediated, because the IgE and tryptase levels were normal. They also noted that future work will be with DTG, TFV, and 3TC.
Real-world evidence on long-acting ART. Levis and colleagues (Abstract 674) presented real-world evidence on CAB-LA/RPV from the OPERA (Observational Pharmacoepidemiology Research and Analysis) cohort. The data are derived from routine medical records and the cohort encompasses more than 100 clinics and approximately 14% of PWH in the US. They reported on 2858 adults (median age, 39 years; 16% women; 42% Black) who received at least 1 dose of CAB-LA/RPV and initiated treatment while being virally suppressed. There were delayed injections in 30% of the cohort and 12% missed at least 1 maintenance injection. However, 95% of the cohort had a plasma HIV-1 RNA level below 50 copies/mL at the last follow-up visit and this did not differ in the subgroup with obesity. Approximately 1% of the cohort experienced confirmed virologic failure. The authors did not identify factors that were statistically associated with confirmed virologic failure; however, there were only 19 confirmed virologic failures in the cohort. The authors concluded that most individuals initiating CAB-LA/RPV were able to remain on this treatment with good virologic success.
Altamirano and colleagues reported on the outcomes for 532 women initiating CAB-LA/RPV within the OPERA cohort (Abstract 676). The outcomes in women appeared similar to those in the full cohort, with the exception of a lower proportion of women receiving all doses on time (44%) compared with the complete cohort. Ninety-four percent of women had a plasma HIV-1 RNA level below 50 copies/mL at the last follow-up visit and confirmed virologic failure (n = 5) occurred in 1.3%. Sax and colleagues presented similar data using the Trio Health cohort (Abstract 675). Among 1198 individuals initiating CAB-LA/RPV, 95% had a plasma HIV-1 RNA level below 50 copies/mL at the last follow-up visit and confirmed virologic failure was observed in 1.6%.
Colesanti and colleagues (Abstract 690) presented data on the outcomes in patients with ongoing viremia who initiated CAB-LA/RPV with and without LEN or ibalizumab. They reported on 81 patients; 2 were LTFU before assessing outcomes. Of the remaining 79 patients, 77 (92%) achieved a plasma HIV-1 RNA level below 50 copies/mL; of these 77 individuals, 2 had ongoing low-level viremia and 2 experienced virologic failure with emergence of resistance. Gistand and colleagues (Abstract 689) presented additional data on CAB-LA/RPV from Ward 86 in San Francisco, California. They reported that 94 of 96 (98%) initiating CAB-LA/RPV while viremic achieved plasma HIV-1 RNA level below 30 copies/mL. Of note, this population experiences frequent housing instability and a high proportion report substance use and mental health issues. These data add to a growing body of evidence for using long-acting ART in those with ongoing viremia.
Other issues with CAB-LA/RPV. Rana and colleagues (Abstract 692) reported additional data from ACTG A5359, a clinical trial of long-acting CAB/RPV in people with adherence challenges. Originally, the protocol required that participants achieve viral suppression and be maintained on oral ART at least 20 weeks prior to being eligible for randomization to CAB-LA/RPV or continuation of oral ART. In subsequent revisions to the protocol, this was decreased to 12 weeks and then to 4 weeks. The definition of viral suppression required for randomization to CAB-LA/RPV was also changed from <50 copies/mL to <200 copies/mL. With these modifications, they found that a higher proportion of people entering the study were able to be randomly assigned to CAB-LA/RPV or to continue oral ART. These modifications also resulted in enrolling more participants with moderate to severe anxiety and moderate to high suicide risk. The authors conclude that the less stringent criteria for initiation of CAB-LA/RPV effectively engaged a more vulnerable population to receive the benefits of CAB-LA/RPV.
Fernandez-Gonzalez and colleagues (Abstract 680) investigated whether an oral lead-in with CAB/RPV impacted pharmacokinetics and long-term outcomes compared with a direct-to-inject strategy using a quasiexperimental design. Use of an oral lead-in did not lead to differing trough concentrations and did not impact virologic outcomes. They found that patient-specific factors such as sex, body mass index (BMI), and smoking status impacted trough concentrations.
Other Antiretroviral Studies
Second-line ART in LMIC. Brown and colleagues (Abstract 669) presented data on the outcomes to second-line ART in South Africa. Using real-world data, they described outcomes of second-line therapy for those failing an EFV-based regimen with 2 plasma HIV-1 RNA levels higher than 1000 copies/mL. They compared responses to LPV boosted by RTV (LPV/r) plus zidovudine (AZT)/XTC compared with DTG/AZT/XTC or DTG/TDF/XTC. They used inverse probability weighting to reduce bias in the choice of treatment to more closely emulate a clinical trial. They reported on 2 primary outcomes at month 24. The 3 groups had similar rates of LTFU or death with 31%, 30%, and 34%, respectively, with no significant differences. They did observe differences in the proportions with plasma HIV-1 RNA levels below 50 copies/mL with the LPV/r group significantly higher than the DTG groups: 51% vs 38% and 38%. The authors concluded that these data support current recommendations for DTG-based second-line therapy and highlight the need for more adherence support for those receiving second-line therapy.
Antiretroviral therapy initiation in those with advanced HIV. Behrens and colleagues (Abstract 658) presented data from an open-label, randomized clinical trial comparing BIC/FTC/TAF and darunavir boosted by cobicistat (DRV/c)/FTC/TAF in those who were naive to ART with advanced HIV, defined as CD4+ count below 100 cells/µL, severe bacterial infection with a CD4+ count below 200 cells/µL, or AIDS or opportunistic infection at any CD4+ count. They enrolled 442 participants (median age, 43 years; 19% women; 62% White; 86% with CD4+ count <100 cells/µL; mean HIV RNA level, 5.6 log10 copies/mL). The primary outcome was a composite of virologic failure or serious clinical events at week 48. This occurred in 24% in the BIC/FTC/TAF group and 34% in the DRV/c/FTC/TAF group (HR, 0.7; 95% CI, 0.48-1.0). This achieved noninferiority of BIC/FTC/TAF compared with DRV/c/FTC/TAF. Virologic outcomes were generally better in the BIC/FTC/TAF group and serious clinical events were similar between groups. Adverse events were generally lower in the BIC/FTC/TAF group. These data provide support for the use of BIC/FTC/TAF in those initiating ART with advanced HIV disease.
Less frequent dosing of BIC/FTC/TAF. Martinez and colleagues (Abstract 659) presented data from a pilot clinical trial investigating less-frequent dosing of BIC/FTC/TAF in those with stable virologic suppression. They randomly assigned 40 participants (10 in each group) to continue daily oral BIC/FTC/TAF, dose BIC/FTC/TAF 3 times per week, 2 doses per week, or weekly dosing. There appeared to be a higher risk of virologic failure when BIC/FTC/TAF was dosed once weekly, with 2 of 10 participants having plasma HIV-1 RNA levels at week 12 and week 48. Resistance-associated mutations did not emerge in any participants and participants resuppressed on daily oral BIC/FTC/TAF.
BIC/FTC/TAF in kidney transplant recipients. Coppinger and colleagues (Abstract 650) presented data on the pharmacokinetics of BIC/FTC/TAF in patients who have undergone renal transplant. This was a single-arm, open-label switch study of PWH who had received a renal transplant. Participants were followed up for 72 weeks and intensive pharmacokinetics were performed 12 weeks after initiating BIC/FTC/TAF. They enrolled 19 participants (68% male; 89% Black; median creatinine clearance, 52 mL/min). Creatinine clearance and tacrolimus concentrations remained stable after the switch of ART to BIC/FTC/TAF. Plasma AUCs of TAF, FTC, and TFV were similar historic controls with similar creatinine clearance levels; however, the intracellular TFV-diphosphate and FTC-triphosphate levels were significantly higher than historic controls with higher creatinine clearance. The BIC AUCs were similar to historic controls with normal renal function. The authors concluded that BIC/FTC/TAF was an appropriate ART regimen for kidney transplant recipients.
Crushed or dissolved BIC/FTC/TAF. It to maintain oral dosing of ART tablets when patients require enteral feeding in the hospital. Mercure and colleagues described the outcomes of 19 patients who received crushed or dissolved BIC/FT/TAF for at least 7 days while receiving enteral feeding. The median duration of the use of crushed or dissolved BIC/FTC/TAF was 19 days. Approximately half of the cohort was viremic at the time of using crushed or dissolved BIC/FTC/TAF. Seventeen (89%) were virally suppressed at follow-up within the next year and 2 required a change of therapy. None developed new resistance-associated mutations. The authors concluded that using crushed or dissolved BIC/FTC/TAF was a reasonable option for maintaining ART during periods of enteral feeding.
Updates on Resistance to Therapies for HIV
Epidemiology of HIV Resistance
Kagan and colleagues (Abstract 730) presented drug resistance epidemiology and trends from 2018 to 2024 based on RNA genotyping of more than 50,000 samples or proviral DNA genotyping of more than 25,000 samples. The cohort was mostly men (76.3% in the RNA group and 78.6% in the DNA group, and the median age was 41 years in the RNA group and 51 years in the DNA group. Interestingly, they found a decrease in resistance to any ART drug class from 2018 to 2024 for RNA and DNA testing (P < .0001). When adjusting the DNA samples for a 20% cutoff, the prevalence of drug resistance mutations was similar in RNA and DNA sequences. They also found dual-class resistance (resistance to nucleoside and nucleotide analogue reverse transcriptase inhibitors [nRTIs] and NNRTIs or InSTIs) declined in RNA and DNA samples from 2018 to 2024. The prevalence of resistance to DOR (the newest NNRTI) and RPV remained stable or slightly decreased from 2018 to 2024, despite the introduction and use of LA RPV. Resistance to CAB also remained stable or slightly decreased from 2018 to 2024. They noted a decrease in prevalence of InSTI mutations in RNA and DNA except for R263K, which increased 3-fold in RNA specimens and 2-fold in DNA specimens from 2018 to 2024. They observed very low rates of K65R; under 1% in 2024 in RNA and DNA specimens. Multidrug class resistance was more likely to be identified in specimens from older adults (those aged 60 to 90 years). The authors concluded that prevalence of RT resistance decreased likely due to using regimens that are fixed dose, better tolerated, and have higher barrier to resistance. Reassuringly, most InSTI-associated resistance remained low and declined except for the R263K mutation, which may impact future InSTI-based regimens.
Armenia and colleagues (Abstract 733) evaluated the epidemiology and frequency of transmitted drug resistance (TDR) among a cohort of 473 newly diagnosed individuals from 2021 to 2024 in Italy. They examined RNA genotyping of these individuals and found a TDR prevalence of 12.3% (when using frequency of ≥20%) and 15.9% (when using frequency of ≥10%). RT TDR was most common, with below 8% for nRTI TDR and NNRTI TDR, 2.5% for PI TDR, and below 0.5% for InSTI TDR. TDR tended to be higher in B-subtype virus than in non-B-subtype virus (P = .031 when using a frequency of 20%). Reassuringly, they found that more than 99.5% of individuals in the cohort were still genotypically susceptible to first-line regimens containing 2 nRTIs and a second-generation InSTI.
D'Antoni and colleagues (Abstract 735) reported on resistance testing for participants in ALLIANCE (Safety and Efficacy of Bictegravir/Emtricitabine/Tenofovir Alafenamide Versus Dolutegravir Plus Emtricitabine/Tenofovir Disoproxil Fumarate in Treatment Naive, HIV-1 and Hepatitis B Co-Infected Adults),9 a study showing that BIC/FTC/TAF was noninferior to DTG plus FTC/TDF in achieving viral suppression in PWH and hepatitis B. Individuals in the ALLIANCE trial were naive to HIV and HBV treatment and underwent genotyping of protease and RT mutations at screening. The researchers conducted genotyping analyses for PIs, RT inhibitors, and InSTIs if patients had virologic failure. At baseline, they observed protease resistance-associated mutations (RAMs) at a rate of 1% to 3% and RT inhibitor RAMs at a rate of 2% to 10%, but these did not affect virologic suppression by the end of the study. In the 14 participants who had virologic failure and did not resuppress, none had treatment-emergent primary RAMs. This study demonstrated the high barrier of resistance of BIC/FTC/TAF against preexisting RT inhibitor mutations.
Christoph (Abstract 741) presented data from a study of veterans with HIV and the impact of resistance on their clinical outcomes. They compared clinical outcomes among 3 groups in a cohort of 7746 individuals: (1) a group with no RAMs; (2) a group with any RAMs; and (3) a group with major RAMs. Mean age and percentage of male sex were similar across the groups. The mean CD4+ count in the no RAMs group was 358 cells/µL, which is higher than the any RAMs group at 343 cells/µL and the major RAMs group at 333 cells/µL (P < .05). The rate of hospitalization was significantly higher for the major RAMs group and the any RAMs group than for those without any RAMs (P < .001). Those with any RAMs were significantly more likely to have any opportunistic infection (5.2% vs 3.8% in the no RAMs group; P = .005). For those who were not taking ART at baseline but then initiated ART during the study period, individuals with no resistance were more likely to continue ART at 1 year than those with any RAMs (P < .0001) and those with major RAMs (P < .0001). Those with major RAMs tended to be the least likely to continue ART at 1 year. Ultimately, resistance appears to be associated with worse clinical outcomes, including increased rates of hospitalizations and opportunistic infections and decreased mean CD4+ cell counts and lower likelihood to continue ART therapy after 1 year.
Resistance to Reverse Transcriptase Inhibitors
Marc and colleagues (Abstract 729) looked at whether nRTI resistance affected viral suppression rates in an open-label trial of virologically suppressed adults on PI-based ART switching to BIC/FTC/TAF. In the 153 participants randomly assigned to switch to the BIC/FTC/TAF arm (vs continuing on their current PI-based regimen), 149 had proviral genotyping available. Fifty-three percent had some nRTI resistance, including 51% with M184V/I, 16.1% with K65R, and 14.1% with M184V/I and K65R. A total of 9.4% had M184V plus at least 3 thymidine analogue mutations (TAMs); 3.4% of the participants had low-level InSTI RAMs (G140R or E138K) and 1 participant had moderate-level InSTI RAMs. They observed that in their primary outcome of virologic failure (HIV viral load ≥200 copies/mL), switching to BIC/FTC/TAF was noninferior to staying on a PI-based regimen. Given the high rates of baseline nRTI resistance in the cohort, the authors conclude that individuals can be safely switched to BIC/FTC/TAF without having prior resistance data available and still achieve virologic suppression.
Soto-Ramirez (Abstract 731) presented epidemiology data of RPV resistance from Mexico, where the NNRTI EFV was part of first-line ART for 10 years. They examined genotyping from 2573 patients who were highly treatment experienced and had experienced at least 2 treatment failures. All patients had previous failure of an NNRTI-based regimen. They found that 26.2% of participants had at least 1 RPV RAM and 14.6% had at least 2 RPV RAMs. The most commonly found RPV RAMs were L100I, K101E, Y188L, and G190A. Of those with RPV mutations, 12.6% had intermediate-level RPV resistance, and 43.2% had high-level RPV resistance. The study authors concluded that in LMICs where genotyping may not be possible and NNRTI-based regimens were first line, pre-existing RPV resistance rates could be high and prohibitive for using long-acting ART.
Abdi and colleagues (Abstract 734) looked at virologic failure in a French cohort of 102 patients who were NNRTI experienced and had previous NNRTI RAMs who were then switched to DOR. Viral suppression was an inclusion criterion; the median duration of viral suppression was 7 years. The median age of patients was 59 years, with a median of 26 years from HIV diagnosis and a median of 22 years of experience with ART. Twenty-five percent of the patients had prior known RAMs to DOR (by RNA genotyping or DNA proviral genotyping), with 64% of these individuals with evidence of major resistance to DOR. In those who continued DOR-based ART at a 2-year follow-up, there were no cases of virologic failure. Of note, the median time since the last NNRTI RAMs were detected was 12 years. The study authors concluded that DOR-based regimens can still be effective in patients who have historic NNRTI resistance who have been virally suppressed for a long period.
Resistance to Integrase Strand Transfer Inhibitors
Hikichi and colleagues (Abstract 155) looked at how nucleocapsid mutations can confer resistance to InSTIs. They observed that these mutations arise in the presence of DTG and can confer moderate resistance to DTG. They found that nucleocapsid mutations can cause the virus to complete integration faster than wild-type viruses. The authors postulate that the shorter time to integration limits the ability of InSTIs to bind and block integration. When they looked at InSTI resistance in HIV-1 virus in peripheral blood mononuclear cells (PBMCs), they observed throughout 40 weeks that the virus sequentially acquired envelope mutations, then nucleocapsid mutations, then InSTI mutations. The envelope mutations allowed increased infection capacity, thereby conferring resistance to InSTIs. They also observed InSTI mutations often traveling with nucleocapsid mutations. They then performed an experiment looking at integrase resistance to DTG, CAB, and RAL with wild-type virus, virus with nucleocapsid mutations, InSTI mutations, and a combination of nucleocapsid and InSTI mutations. The combination of InSTI and nucleocapsid mutations conferred the highest level of resistance; the virus was 11-times more resistant to CAB and 7.5-times more resistant to DTG than wild-type virus. Ultimately, this study found that numerous mutations in other parts outside the integrase genome, including envelope and nucleocapsid, contribute to InSTI resistance.
Penrose and colleagues (Abstract 156) looked at mutations that emerged in individuals in A5381 (Observational Cohort to Assess Therapeutic Efficacy and Emergence of HIV Drug Resistance Following Initiation of Tenofovir-Lamivudine-Dolutegravir [TLD] for First- or Second-Line ART or With Rifampin-Containing TB Treatment: The Hakim Study), specifically mutations emerging outside the integrase genome in individuals after taking TFV/3TC/DTG (TLD). They looked at patients who had confirmed virologic failure on TLD and compared these individuals with matched study entry samples with viremia before TLD exposure. They used HIV genome sequencing on these plasma samples to determine which mutations arose after exposure to TLD. They found 10 participants who had virologic failure after exposure to TLD who developed nucleocapsid mutations (frequency ≥ 20%) that were not found in the pre-TLD samples; 6 of these participants had nucleocapsid mutations at a frequency of 50% or higher. Of note, they observed 2 mutations, N27I and M46I, that were selected for in vitro in the previous experiment conducted by Dr Hikichi and colleagues. One participant had a major integrase drug resistance mutation as well as nucleocapsid mutations. They looked at phenotypic assays with these mutations to assess DTG resistance; N27I and M46I alone conferred 1.8- to 2.1- fold resistance to DTG. When N27I was combined with an integrase RAM (R263K) observed in 1 participant, it caused more than a 5-fold resistance to DTG. The authors concluded that these are the first clinical data to show that nucleocapsid mutations can confer resistance to InSTIs, can emerge with or without integrase mutations, and can have an additive effect in conferring resistance to InSTIs.
Choudhuri and colleagues (Abstract 722) examined why CAB appears to be significantly more susceptible to combination integrase mutations such as E138K/G140A/Q148R over other InSTIs, such as DTG. They used 50% maximal effective concentration (EC50) measurements, high-resolution electron microscopy structures, and computational modeling. They observed that the triple integrase mutant causes disruptions in the active integrase site, decreasing binding to InSTIs, and that CAB is particularly susceptible to this due to molecular dynamics, compared with DTG. The ability of DTG to resist these disruptions allows it to remain more potent against these triple mutants.
Hansen and colleagues (Abstract 724) examined in vitro resistance to GS-1720, a once-weekly oral InSTI currently in development. They tested the antiviral activity of GS-1720 against HIV-1 viruses with InSTI RAMs, including G140A/S and Q148H/K. They found that GS-1720 was still potent (which they defined as EC50 change of <10-fold) against 91% of the isolates, comparable with BIC. They reported that high-level resistance against GS-1720 (defined as EC50 change of >10-fold) required at least 3 integrase mutations, including G140A/S + Q148H/K + L74I, T97A, or E138A/K. GS-1720 was still active against strains with T97A. This represents a novel LA InSTI that appears to be as robust as BIC.
Calvez and colleagues (Abstract 732) looked at the frequency and epidemiology of integrase mutations in France from 2007 to 2024 with the emergence of various members of the InSTI class. Their cohort included 3100 people in 2007, which increased to 4364 in 2024. Overall InSTI use increased dramatically from 3.3% in 2008 to 71.2% in 2024. They observed an increase in InSTI resistance within the first 3 years of the cohort, but this decreased over time from 44.8% in 2008 to 2.7% in 2023 for DTG, BIC, and CAB. Resistance to firstgeneration InSTIs remained higher in 2024, with 8.5% prevalence of resistance to RAL and 9.8% rate of resistance to elvitegravir. The authors postulate that the high initial rate of InSTI resistance observed was likely due to RAL paired with partially inactive agents, but with the use of newer agents with an increased barrier to resistance, this has prevented continued emergence of InSTI resistance.
Merad and colleagues (Abstract 737) presented data from a cohort of virologically suppressed patients on InSTI-based regimens in France to see if proviral geno-typing showing InSTI mutations was clinically relevant. They included 26 patients who had at least 1 mutation conferring resistance to the entire InSTI class (G118R, E138A/K, G140S, N155H, or R263K). Most patients were on a second-generation InSTI-based regimen, including 1 participant on CAB. In 69% of these participants, DRMs were found in defective proviruses (defective proviruses were defined as those with stop codons or APOBEC-mediated G-to-A hypermutation). Ultimately, none of these individuals had virologic failure on their InSTI-based regimens, suggesting that archived InSTI mutations may not predict virologic failure on InSTI-based regimens, especially as many of these mutations are associated with defective proviruses. This is consistent with a recent article suggesting that mutations caused by APOBEC-mediated G-to-A hypermutation should be excluded from next-generation sequencing reporting.10
Tumiotto (Abstract 739) presented data from a French ARNS (National Agency for Research on AIDS and Viral Hepatitis) cohort looking at virologic failure after switching to BIC/FTC/TAF between 2018 and 2021 and the emergence of resistance. They observed 10.4% (66 of 636) of participants with virologic failure (defined as 2 consecutive viral loads >50 copies/mL or 1 viral load >200 copies/mL), although 71.4% of the cohort had suppression on BIC/FTC/TAF. More than half were unsuppressed at baseline, with an HIV RNA viral load level above 50 copies/mL. They saw a 9.1% (6 out of 66) rate of resistance to any component of BIC/FTC/TAF during the study period, with M184V being the most common in 5 of 6 participants, along with 1 N155H, and 1 R263K. The authors concluded that these data demonstrated that BIC/FTC/TAF is effective in achieving virologic suppression and is associated with low rates of emergence of InSTI resistance.
Martin and colleagues (Abstract 742) evaluated resistance in a cohort in Uganda from 2012 and 2022, specifically choosing this study period as it was before and after when DTG was recommended as first-line ART in 2018. They considered all viremic patients (HIV RNA level ≥ 1000 copies/mL) and their genotypic RNA level resistance testing results. Between 2014 and 2022, there was an increase in viral suppression rate in the cohort to 90% (95% CI, 88-90), an increase of 1.58-fold across the study period. In the baseline genotyping results before treatment initiation, they observed stable prevalence of resistance to nRTIs, PIs, and InSTIs, but an increase in resistance to NNRTIs from 5% (95% CI, 4-7) in 2012 to 14% (95% CI, 9-21) in 2022. The most commonly observed NNRTI mutations were K103N and E138A. The overall prevalence of resistance to any class of ART decreased from 47% (95% CI, 38-59) in 2014 to 29% (95% CI, 22-37) in 2022 in treatment-experienced individuals with viremia. The rate of DTG emergent resistance was 1.85% in 2022 in this group, after the rollout of DTG-based therapy. There was an increase in the S153Y InSTI mutations in 2021 but most individuals with this mutation had TDR that was genetically linked to another person with HIV with the S153Y mutation. The authors conclude that the overall prevalence of resistance, including resistance to InSTIs, remains low after the rollout of DTG-based therapy but continued surveillance is needed, in particular for NNRTI resistance and the S153Y mutation, especially as LA ART is being used more frequently.
Two abstracts covered the use of plasma DTG drug levels and their association with the absence of DTG resistance. Steegen and colleagues (Abstract 743) performed a study in adults in South Africa who were on a DTG-based regimen and had a detectable viral load. In those with detectable DTG levels, 13% had DTG resistance. However, 93% of individuals with undetectable DTG levels had no DTG resistance; the authors concluded that consistent undetectable DTG levels had a high negative predictive value of 97.7% for DTG resistance and could be a useful predictive tool. Coetser and colleagues (Abstract 744) also looked at plasma DTG levels as a predictive tool for DTG resistance in a cohort of viremic South African adults with viremia (HIV-RNA level >400 copies/mL) on a DTG-based regimen between 2019 and 2022. Those with detectable DTG levels were more likely to have DTG resistance, with an odds ratio of 9.54 (P = .0001), and undetectable plasma DTG levels had a very high negative predictive value of 94.2% (95% CI, 85.1-97.9). These 2 abstracts demonstrate that plasma DTG levels may be a useful and economical tool in predicting DTG resistance, especially in LIMCs.
McCluskey and colleagues (Abstract 745) evaluated the association between nRTI resistance and DTG resistance when switching to a TLD-based regimen in sub-Saharan Africa. Three percent of 999 participants switching from an NNRTI-based regimen to TLD had levels of more than 500 HIV RNA copies/mL. None of the participants who had baseline nRTI resistance or nRTI emergent resistance developed resistance to DTG after 1 year on TLD.
Resistance to Lenacapavir
Bartolini and colleagues (Abstract 721) examined the genetic barrier to resistance of LEN based on subtype-B and non-B HIV subtypes as well as exposure to ART. In a cohort of viruses from their registry in Italy, there was no difference in susceptibility to LEN between treatmentnaive individuals and highly treatment-experienced individuals (P = .101) or between subtype-B viruses and non-B-subtype viruses (P = .0514). There was no difference in time to viral breakthrough of viruses exposed to between treatment-naive individuals and highly treatment-experienced individuals (P = .0284). However, the time to viral breakthrough with LEN was significantly lower with subtype-B viruses than with non-B-subtype virus (P = .047). This finding is at least reassuring that heavily treatment-experienced individuals (who may be more likely to be on LEN) are not more likely to have resistance to LEN than treatment-naive individuals.
McFadden and colleagues (Abstract 727) modeled ways to overcome LEN resistance with the known capsid mutation N74D. Modifications made at the binding pocket of N74D-carrying viruses, where LEN binds to the capsid protein, allowed improved potency of LEN. Changes to the binding pocket may represent a future target or strategy in overcoming LEN resistance.
Vanderveen and colleagues (Abstract 736) presented resistance analysis data from a phase II trial of once-weekly ISL plus LEN in PWH who are virologically suppressed. Participants underwent proviral DNA genotyping at baseline, and those with primary RT inhibitor RAMs by screen or historic genotype were excluded from the study. Five participants who were enrolled had RAMs in their screening genotype but continued in the study, including 1 participant in the ISL plus LEN arm who had the M184V mutation. All these participants maintained virologic suppression by week 48. One participant in the study had virologic failure and underwent post baseline resistance analysis but had no treatment-emergent drug resistance.
Margot and colleagues (Abstract 738) presented resistance analysis data from ARTISTRY-1 (Study to Compare Bictegravir/Lenacapavir Versus Current Therapy in People With HIV-1 Who Are Successfully Treated With a Complicated Regimen), a phase II/III trial examining the efficacy of LEN plus BIC vs a complex multitablet ART regimen. They examined genotyping at baseline and at week 48 for participants who had an HIV RNA level of 200 copies/mL or higher. Eighty-one percent of participants had nRTI resistance, 65% had NNRTI resistance, and 46% had PI resistance at baseline. Eleven percent of participants at baseline had primary InSTI RAMs, but these were found on proviral genotyping only. One participant at week 48 had virologic failure requiring genotyping and had no InSTI-related resistance but had the N74 capsid mutation conferring some resistance to LEN; this participant had known RT and PI resistance and had resuppression on LEN plus twice-daily DTG plus maraviroc. These data support the use of LEN plus BIC in a treatment-experienced population with high rates of baseline multiclass resistance.
Advances in Maternal and Pediatric Health
Couples-Focused Interventions for Improving Partner and Couples HIV Testing and Counseling
Darbes and colleagues (Abstract 118) presented findings from a randomized clinical trial assessing the efficacy of a home-based intervention designed to improve couples HIV testing and counseling (CHTC) and family health during pregnancy and postpartum in Kenya. The study was based on the premise that engaging pregnant women and their male partners can improve decisionmaking, healthcare utilization, and maternal and child health outcomes. Conducted between 2019 and 2024, the trial recruited 800 pregnant women, two-thirds of whom had HIV, during antenatal care, with their male partners subsequently contacted. The couples were randomly assigned to 1 of 3 groups: home visit intervention (HV; n = 267); HIV self-test kits (HIVST; n = 266); or standard of care (SOC; n = 267). The HV group received 5 home visits, with 2 during pregnancy and 3 during the postpartum period, from pairs of male and female lay health workers; these sessions involved health education and couple relationship skills training, along with CHTC. The HIVST group was provided with self-test kits for the couples to use independently. Interviews with partners were conducted during pregnancy and at 3 and 12 months postpartum, with HIV RNA results obtained from medical records. Couples in the HV group were significantly more likely to engage in CHTC than those in the SOC group (adjusted risk ratio [aRR], 4.22); the HIVST group also demonstrated increased engagement in CHTC (aRR, 3.69). Home visits led to an increase in the number of HIV tests conducted, resulting in more new HIV diagnoses in men and identification of serodiscordant couples. In addition, positive relationship dynamics, including relationship satisfaction, dyadic trust, and communication between couples, were significantly greater in the HV group. Maternal and child health care utilization, such as antenatal care visits and postpartum and postnatal care, was comparable across all groups. WWH in the HV group had a higher likelihood of achieving viral suppression at 18 months (aRR, 1.07). In terms of outcomes for HIV-exposed infants at 18 months, there were 2 infant HIV infections diagnosed, with 1 in the HV group and 1 in the HIVST group. Overall, the study showed that the home-based couples intervention led to improvements in maternal viral suppression rates, increased couple HIV testing, and enhanced relationship quality. The authors concluded that engaging the partners during pregnancy and postpartum is a promising strategy for HIV prevention and family health promotion.
Kwena and colleagues (Abstract 1181) presented an analysis evaluating moderators of a home-based couples intervention on the uptake of CHTC in Kenya, based on a randomized clinical trial involving 800 couples. Half of the participants (50.0%) were aged 25 to 34 years; 40.6% of the couples experienced an age disparity of 5 to 10 years; and 22.7% had a disparity greater than 10 years. The median relationship duration among the couples was 4 years, and the median parity was 2. Of the 9 factors examined, including relationship duration, stigma perceptions, male partner age disparity, wealth (measured by an index score), education, couple communication, relationship satisfaction, women's HIV status, and parity, the home-based intervention showed a significantly better beneficial impact on CHTC uptake in couples where the woman was HIV negative at baseline and had lower parity. The authors suggested that future scale-up of the CHTC intervention should prioritize HIV-negative women with lower parity.
Chagomerana and colleagues (Abstract 1180) investigated the impact of passive partner notification combined with oral HIV self-test kits for male partners of pregnant women. The study was an open-label, randomized trial involving 200 HIV-negative pregnant women receiving antenatal care in Malawi. The women were randomly assigned to receive standard of care, which included partner notification inviting their partners for HIV testing, or the intervention, which included partner notification along with 1 HIV self-test kit for home testing by their partners. The study was later modified in response to changes in national guidelines with rollout of HIV self-tests for partners of pregnant women. As a result, women in the SOC group received partner notification along with 1 HIV self-test kit, and those in the intervention group received partner notification and 2 HIV self-test kits. Overall, regardless of the intervention, only 70 (35%) of male partners reported to the clinic for confirmatory HIV testing, and all men who underwent clinic-based testing had previously tested negative at home. The authors found that the inclusion of the HIV self-test kit as part of passive partner notification did not improve clinic-based HIV testing among male partners of pregnant women within 30 days of study enrollment. They emphasized the need for novel strategies to effectively reach and engage male partners in HIV testing and to enhance awareness of their HIV status.
Hampanda and colleagues (Abstract 991) evaluated the efficacy of an adapted couples-based counseling intervention compared with enhanced SOC in improving maternal postpartum viral suppression rates and other health outcomes, including ART adherence, pediatric HIV testing, and exclusive breastfeeding, in Zambia. The intervention focused on enhancing relationship skills such as trust, support, and communication through 3 counseling sessions. Although no significant differences were observed between the study groups in primary or secondary outcomes, high rates of viral suppression at baseline and at 6-month follow-up were noted among postpartum women enrolled with their male partners. In addition, there were high rates of daily ART adherence, pediatric HIV testing, and exclusive breastfeeding.
ART Updates in Women, Including Interactions With Hormonal Contraception
Ryan and colleagues (Abstract 119) presented pharmacokinetic data on drug-drug interactions between 2 hormonal contraceptives, levonorgestrel (LNG) implants and depot medroxyprogesterone acetate (DMPA) injectable, and DTG-based ART among WWH. Conducted in Botswana, this phase IV, open-label, parallel-arm, nonrandomized pharmacokinetic study enrolled 140 women, including 70 WWH on DTG-based ART and 70 ART-naive, HIV-negative women. Among these participants, 70 women (35 WWH, 35 HIV negative) chose to start LNG, and the other 70 (35 WWH, 35 HIV negative) opted for DMPA for contraception. Results showed no statistically significant differences in the mean AUCs over 24 weeks for LNG between WWH on DTG-based ART (14,899 pg.week/mL [mean cumulative concentration in picograms/mL over a 24-week period]) and HIV-negative women (12,039 pg.week/mL), with a geometric mean ratio (GMR) of 1.30. The mean AUC over 12 weeks for PMPA injectable was also similar between the groups (10,887 pg.week/mL for WWH vs 10,201 pg.week/mL for HIV-negative women; GMR, 1.10). Concentrations of DTG were consistent across the 2 arms, with no statistically significant differences in median DTG concentrations before and after contraceptive use. The contraceptive methods were well tolerated, with no unintended pregnancies or serious adverse events reported. The authors concluded that there were no clinically significant drug-drug interactions between DTG-based ART and either the LNG implant or DMPA injectable and confirmed the contraceptives to be safe and effective options for WWH on DTG-based ART.
Levis and colleagues (Abstract 676) demonstrated the high effectiveness of CAB-LA/RPV in treatment-experienced WWH. This analysis, based on the OPERA cohort study, utilized clinical data from EHRs across 101 clinics in 23 US states and territories. The study evaluated clinical outcomes in 415 treatment-experienced women who were virally suppressed at baseline and initiated CAB-LA/RPV monthly or bimonthly. The cohort was racially and ethnically diverse, with 64% identifying as Black and 20% as Hispanic. Among the 381 women who completed the initiation of CAB-LA/RPV and had at least 1 viral load test during follow-up, the majority maintained viral suppression (94% at the last followup visit and 82% at all follow-up visits). Confirmed virologic failure was rare, occurring in fewer than 5 women (≤1.3%).
ART Pharmacokinetics and Safety in Infants and Children
DTG is currently not recommended for neonates due to the absence of dosing guidelines and safety data. The PETITE-DTG (Pharmacokinetics and Safety of Dolutegravlr in Neonate) trial (Abstract 122) is an ongoing phase I/II, open-label, 2-stage study in South Africa aimed at evaluating the multidose pharmacokinetics and safety of 2 available pediatric DTG formulations for use in neonates: a scored dispersible tablet (DTG-DT, 10 mg, used as a half-tablet of 5 mg) and an oral dispersible film (DTG-OF, 5 mg). The study enrolled 43 term neonates exposed to HIV, with birth weights of 2 kg or more, and born to mothers on DTG-based ART. The trial comprised 2 stages: stage 1 provided data on a single dose of DTG-DT that informed the multidose strategy for stage 2, in which neonates received 5 mg DTG every 48 hours for the first 2 weeks of life, followed by daily doses through week 4 of life. Neonates were randomly assigned to receive DTG-DT or DTG-OF in addition to zidovudine (ZDV) for postnatal prophylaxis. Results from 38 neonates (18 who initiated DTG-DT and 20 who initiated DTG-OF) in stage 2 showed that the formulations achieved the predefined target DTG concentration of geometric mean (GM) Ctau of 0.67 µg/mL or more; all neonates had DTG Cmax less than 17.0 µg/mL. There were no adverse events that were related to DTG, although 2 unrelated grade 3 adverse events were reported in 1 neonate. All neonates tested negative for HIV at the end of the study. The study showed that the multidose DTG dosing schedule was safe and achieved pharmacokinetic targets for the DTG formulations, providing the first evidence to support the expansion of DTG dosing recommendations for neonates from birth through the first 4 weeks of life.
Viljoen and colleagues (Abstract 1045) reported on the acceptability of the 2 pediatric formulations of DTG, a dispersible tablet and an oral film, among 29 mothers with HIV whose neonates were enrolled in the PETITE-DTG study. The mothers expressed high levels of acceptability for the DTG formulations, despite initial reservations about the film and some difficulties with the preparation and administration of the dispersible tablet. The authors emphasized the importance of peer support and engagement with mothers and health care workers, who are the end-users of these novel DTG formulations, as a way to improve their uptake.
Archary and colleagues (Poster 1046) presented 12-week interim data from children weighing between 20 and 40 kg as part of IMPAACT 2036/CRAYON (Cabotegravir and Rilpivirine Long-Acting Injections in Young Children), a phase I/II, multicenter, open-label, noncomparative study designed to evaluate the safety, tolerability, acceptability, and pharmacokinetics of oral CAB/RPV and intramuscular CAB-LA/RPV in children aged 2 years to less than 12 years with HIV. A total of 34 children contributed to the pharmacokinetic analysis, and 20 children who completed treatment through week 12 contributed to the safety data. The median age of the children was 10 years and the median weight was 26.2 kg. All children maintained viral suppression through 12 weeks. The authors reported that administering CAB-LA/RPV every 4 weeks, following a 4-week oral lead-in, resulted in drug exposures comparable with those observed in adults and adolescents. Injection-site pain was the most reported adverse event; no new or unexpected safety issues were identified.
In Abstract 1047, pharmacokinetic and safety data for DTG dosing using a 5 mg dispersible tablet formulation were presented for 18 neonates exposed to HIV, including 11 infants whose mothers received DTG and 7 infants without maternal DTG exposure. This analysis was part of IMPAACT 2023 (A Phase I Study of the Safety, Tolerability, and Pharmacokinetics of Dolutegravir in Neonates Exposed to HIV-1), a study examining the pharmacokinetic, safety, and tolerability of DTG in singleton, full-term, HIV-exposed infants during their first 4 to 6 weeks of life. The 5 mg dispersible tablet was given every other day until day 14 of life, followed by once-daily dosing; the infants received their first DTG dose between 0 and 5 days of age. Intensive pharmacokinetic sampling occurred at 7 days and 4 weeks following the initiation of DTG. DTG exposures were evaluated using protocol-defined target GM exposures: Ctrough above 0.697 μg/mL, AUC0-tau above 37 μg*h/mL, and Cmax below 18.35 μg/mL. No clinically significant differences were observed between neonates with maternal DTG exposure and those without. The findings showed that DTG exposures met the protocol-defined target GM exposures; however, there was notable variability, with 5 of the 18 neonates having trough concentrations below the target at 7 days post initial dose, and 1 neonate having a trough concentration below the target at week 4. The DTG formulation was well tolerated, with no unexpected adverse events reported among the neonates. The authors indicated that the evaluation of this DTG dosing regimen, including the investigation of potential sources of PK variability, is ongoing.
Panjasawatwong and colleagues (Abstract 1048) presented results from a pharmacokinetic modeling and simulation study to evaluate whether administration of dispersible abacavir (ABC)/3TC (30/15 mg) tablet (one-quarter tablet) every 48 hours during the 2 weeks of life would provide adequate exposures in term neonates exposed to HIV. This less-frequent dosing schedule would align with the newly proposed DTG dosing every 24 to 48 hours, thereby simplifying the treatment regimen. Of the 24 term neonates included in the study, the median body weight was 3125 g, and the median postnatal age was 8 days at the time of the first pharmacokinetic visit. For ABC, the simulations predicted that the GM AUC0-48 remained within target exposure levels with the every-48-hour dosing compared with the every-24-hour dosing during the first 2 weeks of life. Similarly, for 3TC, the GM AUC0-48 remained with the target with every-48-hour dosing. The predicted plasma exposures of ABC and 3TC during the 2 weeks of life in the neonates were found to be comparable with young children.
Mujuru and colleagues (Abstract 1049) presented safety and efficacy outcomes at 24 weeks of TAF/FTC in combination with cobicistat-boosted (c) PIs (atazanavir [ATV]/c) or DRV/c) in children aged 2 to younger than 12 years and weighing 14 to under 40 kg. This analysis was part of the GS-US-216-0128 trial, an ongoing phase II/III, multicenter, open-label, multicohort trial evaluating TAF/FTC and boosted PIs in children and adolescents with HIV. The analysis focused on 2 specific cohorts: children who were aged 6 years to younger than 12 years weighing 25 to under 40 kg (cohort 2), and children aged 2 years and older and weighing 14 to less than 25 kg (cohort 3). The majority of participants maintained or achieved viral suppression through 24 weeks of treatment, with rates of 96% in cohort 2 and 92% in cohort 3. In addition, TAF/FTC in combination with ATV/c or DRV/c was well tolerated. No clinically significant metabolic, bone, and renal changes from baseline to week 24 were observed. These findings support the need for further evaluation of TAF/FTC in combination with ATV/c or DRV/c in children with HIV, including extended 48-week follow-up data.
Genotypic Resistance in Children Receiving Second-Line Antiretroviral Therapy
Bamford and colleagues (Abstract 123) presented data on genotypic resistance in 919 children from the pediatric CHAPAS-4 (Children With HIV in Africa - Pharmacokinetics and Acceptability of Simple Second-Line Antiretroviral Regimens) trial,11-13 which took place in Zambia, Zimbabwe, and Uganda. This open-label, factorial randomized trial demonstrated superior virologic efficacy at 96 weeks for TAF plus FTC compared with SOC (ABC or ZDV plus 3TC), and for DTG compared with LPV/r and ATV/r in children aged 3 to 15 years initiating second-line ART following treatment failure of NNRTI-based first-line ART. Among the 713 children with baseline resistance testing, 665 (93%) had an M184V/I mutation. At baseline, intermediate- to high-level resistance to the allocated nRTI was observed in 30% of those receiving ABC, 10% receiving ZDV, and 15% receiving TAF, supporting the use of TAF in second-line ART. At 96 weeks, 124 of 908 children (14%) had a viral load of 400 copies/mL or higher, with more than half of these children (54%; 67/124) undergoing RT sequencing at baseline and week 96. No new intermediate- to high-level resistance to the allocated nRTI was observed. Three children (2 on DTG/ZDV/3TC and 1 on LPV/r/ZDV/3TC) developed high-level M184/V resistance. Intermediate-level/high-level resistance to anchor drugs (LPV/r, ATV/r, DRV/r, or DTG) at 96 weeks was infrequent, with only 3 children developing resistance (1 to ATV/r and 2 to DTG; 2 children with DTG resistance had received ZDV/3TC). The findings from this study will help inform second-line treatment guidelines for children with HIV.
Weight Gain and Cardiometabolic Dysfunction in Youth With Perinatally Acquired HIV and in Women With HIV
Kahts and colleagues (Abstract 811) investigated sex differences in myocardial steatosis among youth with perinatally acquired HIV as part of the CTAAC (Cape Town Adolescent Antiretroviral Cohort)-Heart study in South Africa, which examined myocardial fibrosis and inflammation using cardiovascular magnetic resonance (CMR). The substudy used proton cardiac magnetic resonance proton spectroscopy (1H-cMRS) to evaluate myocardial triglyceride (mTG) content in 72 youths with perinatally acquired HIV. The results showed that women had a higher BMI (24 vs 20 kg/m2) and waist-to-height ratio (0.48 vs 0.43) than males, along with elevated mTG content (0.34% vs 0.13%) and a greater prevalence of pathologically elevated mTG levels (defined as >0.5%; 42% vs 17%). Even after adjusting for age, BMI, HIV viral load, and InSTI-based ART, women had significantly higher odds of elevated mTG content. These findings suggest that youth with perinatally acquired HIV, particularly women, may be at increased risk for myocardial steatosis and fibrosis, highlighting the importance of early screening, prevention, and potential interventions.
Yang and colleagues (Abstract 895) investigated the metabolic signatures associated with weight gain in WWH who initiated InSTI-based ART in the WIHS (Women's Interagency HIV Study) cohort. The study compared weight changes among WWH who were virally suppressed on ART and switched to or added InSTI (n = 38) with those who continued on non-InSTI-based ART (n = 88) and those without HIV (n = 66). The study found that, compared with women on non-InSTI-based ART and women without HIV, metabolic pathways related to polyunsaturated fatty acids, such as linoleic acid, arachidonic acid, and eicosanpentaenoic acid, were associated with weight gain in the InSTI group. Furthermore, inflammatory polyunsaturated fatty acids and oxylipins were enriched in these pathways. These findings indicate that certain metabolic profiles may predispose WWH to weight gain after starting InSTI-based ART. The results emphasize the need for additional research to understand the biologic mechanisms underlying InSTI-related weight gain, particularly in women.
Abelman and colleagues (Abstract 903) explored the association between HIV, menopause, and insulin resistance in women using data from the WIHS cohort. It specifically examined whether HIV and menopause are linked to increases in insulin resistance and whether switching to an InSTI impacts this association. Based on data from 389 nonpregnant WWH who were virally suppressed on ART (including 133 women who switched to an InSTI and 256 who did not) and 334 women without HIV, WWH experienced faster annual increases in insulin resistance during late perimenopause and menopause than women without HIV; this association persisted even after adjusting for waist circumference. However, switching to an InSTI did not lead to an acceleration in insulin resistance. The results suggest that there may be additional mechanisms affecting insulin resistance in WWH beyond visceral adiposity and highlight the importance of diabetes screening and prevention in middle-aged WWH.
Goedecke and colleagues (Abstract 906) examined insulin dynamics in early pregnancy in 894 pregnant women who were at 14 weeks of gestation or less. The study included 357 WWH receiving ART consisting of TDF/3TC/DTG and 537 HIV-negative women. The research was part of the ORCHID (Obesogenic Origins of Maternal and Child Metabolic Health Involving Dolutegravir) study conducted in South Africa. The women completed a 75 g oral glucose tolerance test at enrollment to assess insulin sensitivity, first-phase insulin response, beta-cell function, and measures of beta-cell function (glucose sensitivity) and postprandial insulin. WWH on TFV/3TC/DTG had higher insulin sensitivity and beta-cell function than HIV-negative women, with longer use of TFV/3TC/DTG (more than 90 days) linked to enhanced beta-cell function. The authors concluded that pregnant WWH receiving TFV/3TC/DTG may have a reduced risk of developing diabetes in early pregnancy compared with HIV-negative women and stressed the need for additional longitudinal research to investigate whether this metabolic risk profile persists throughout pregnancy and postpartum.
Using data from the ORCHID study, Shuaib and colleagues (Abstract 984) evaluated patterns of adiposity and gestational weight gain in pregnant South African WWH and receiving TFV/3TC/DTG compared with HIV-negative pregnant women, focusing on gestational weight gain and body composition. Among 949 women, WWH experienced a lower rate of gestational weight gain compared with HIV-negative women, with no significant differences in patterns of adiposity based on HIV status or duration of DTG use. These findings showed that DTG is not associated with excess gestational weight gain in pregnant South African women.
Meyer and colleagues (Abstract 1000) presented additional findings from a secondary data analysis of the ORCHID study, which examined the impact of TFV/3TC/DTG use during pregnancy and maternal BMI on adverse birth outcomes. The authors did not find any association between HIV status and the use of TFV/3TC/DTG during pregnancy with adverse birth outcomes. Maternal BMI appeared to be a factor influencing birth outcomes in this cohort, as a higher maternal BMI, independent of HIV and ART use, was associated with a decrease in preterm delivery, low birth weight (<2500 g), and small for gestational age (<10th percentile), as well as an increase in high birth weight (>4000 g) and large for gestational age (>90th percentile).
In the first genome-wide association study, Jensen and colleagues (Abstract 985) assessed genetic risk factors for weight change from delivery to 48 weeks postpartum in 159 WWH who initiated EFV- or DTG-based ART during late pregnancy as part of the DOLPHIN-2 (Dolutegravir Versus Efavirenz in Women Starting HIV Therapy in Late Pregnancy) study. The authors identified SLC47A2, which encodes the multidrug and toxin extrusion transporter 2-K (MATE2-K), as a potential genetic marker for weight gain in this population. This gene has previously been associated with interindividual variability in serum creatinine level changes following the initiation of DTG.
Health Outcomes of Children With In-Utero HIV and ART Exposure
A Themed Poster Discussion focused on the impact on growth, obesity, and cardiometabolic outcomes in children with HIV exposure who remained HIV uninfected. Tobin and colleagues (Abstract 1021) examined the gut microbiota of children who were exposed to but not infected with HIV (CHEU) compared with that of children unexposed to HIV (CHU) to evaluate the potential impact of specific microbes, particularly Bifidobacterium, on growth outcomes. Conducted in Kenya, the study involved 16S rRNA sequencing on rectal swabs from 319 children (164 CHEU and 155 CHU) at 3 months of age, as part of a prospective observational cohort study of 350 pregnant women with and without HIV and their infants. Overall, 90% of WWH received DTG-based ART, with 75% starting ART before pregnancy and 65% receiving cotrimoxazole prophylaxis. Among CHEU, 82% began antiretroviral prophylaxis at birth, and 98% received cotrimoxazole prophylaxis starting at 6 weeks of age. The findings indicated that CHEU had less weight gain (as measured by weight-for-age [WAZ] scores) than CHU, whereas HIV exposure had a minimal effect on overall microbiome composition at 3 months of age. No significant differences in gut microbiome diversity or genus-level abundances were detected between the 2 groups. However, a stronger association was noted between Bifidobacterium relative abundance and WAZ at 3 months of age in CHEU (P = .06; rho, 0.1764), with the difference in weight gain in CHEU influenced by maternal cotrimoxazole prophylaxis. The authors suggested that maintaining Bifidobacterium populations in the gut may be important for promoting growth in these children.
The biologic mechanisms underlying the increased risk of impaired growth and development, or stunting, in CHEU are not fully understood. Kakayi and colleagues (Abstract 1022) examined the role of insulin-like growth factors (IGFs) in stunting among CHEU by analyzing serum samples from 213 infants enrolled in the IMPAACT P1084s (Study of Effects of Tenofovir on Bone Health and Kidneys During Pregnancy and Breastfeeding) substudy in Uganda, South Africa, and Malawi. The study measured insulin-like growth factor (IGF)-1, IGF binding protein (IGFBP)-1, and IGFBP-3 using validated assays. Among the infants, 47% were male, with a median weight within 5 days of birth of 2,9 kg and length of 48 cm, resulting in a median length-for-age Z (LAZ) score of -1 SD. Higher IGF-1 levels at birth were significantly associated with a reduced risk of stunting at 26 weeks, but IGFBP-1 and IGFBP-3 did not show similar associations. At 74 weeks, increased IGF-1 was associated with improved LAZ, whereas higher IGFBP-1 was linked to decreased LAZ, and IGFBP-3 was associated with increased LAZ. Although growth factor levels were associated with LAZ at later ages, the modest associations observed at birth and future stunting indicate that variations in IGF levels may not be the main determinants influencing stunting in CHEU. The authors highlighted the need to investigate other mechanisms beyond the growth hormone/IGF axis in future research.
Dogon and colleagues (Abstract 1023) investigated the association between maternal gestational lipid biomarkers and cardiometabolic risk factors in CHEUs and CHUs enrolled in the Drakenstein Child Health Study, a longitudinal birth cohort in South Africa. The substudy randomly sampled 260 mother-child pairs, consisting of 100 CHEUs and 160 CHUs. Using multivariable linear regression, the researchers assessed the associations between maternal lipid biomarkers collected during the second trimester (20-28 weeks gestation) and child cardiometabolic indicators at ages 5 to 8 years. The study showed significant associations between maternal total cholesterol (TC) and child TC and low-density lipoprotein (LDL) levels, with a stronger positive correlation observed in CHEUs than in CHUs. In addition, maternal high-density lipoprotein (HDL) levels were positively associated with child HDL levels, although this association was only statistically significant in CHUs. No significant associations were identified between maternal lipid levels and other cardiometabolic indicators, such as hemoglobin A1c, blood pressure, and body composition. The findings indicate that optimizing maternal lipid profiles during pregnancy could improve cardiometabolic health outcomes in children, especially those exposed to HIV in utero, and warrant further investigation in larger cohort studies.
The global scale-up of prenatal ART has successfully reduced mother-to-child transmission of HIV to less than 1%, resulting in more than 20 million individuals worldwide with perinatal HIV exposure without infection (PHEU). Evidence is accumulating that infants and children with PHEU face increased metabolic complications; however, the long-term sequelae of HIV exposure into adulthood remain unclear. In a prospective observational study, Chahal and colleagues (Abstract 1024) evaluated 41 young adults with PHEU and 41 matched controls without HIV exposure or infection, with an average age of 24 years (46% young men). The study found that young adults with PHEU had a higher BMI (effect size, 5.1 kg/m2) and greater rates of class 2 or 3 obesity (41% vs 5%). In addition, they exhibited higher rates of impaired glucose tolerance (21% vs 3%), hypertension (20% vs 0%), and hepatic steatosis (44% vs 15%), resulting in a higher composite metabolic comorbidity score. PHEU status remained significantly associated with BMI and composite comorbidity score even after adjusting for maternal factors (such as BMI and age at delivery) and young adult factors (such as food insecurity and adverse childhood experiences). Notably, 89% of those with obesity and 31% of those without obesity in the PHEU group had at least 1 comorbidity. Furthermore, maternal prenatal CD4+ cell count was inversely associated with cardiovascular risk and hemoglobin A1c level, suggesting that maternal immune dysregulation may contribute to the increased risk of obesity and metabolic complications in PHEU individuals. Overall, this study, the largest to date involving the oldest cohort, found that young adults with PHEU experience a high burden of obesity and metabolic comorbidities. Future epigenetic and proteomic studies are warranted to further explore the mechanisms underlying metabolic disease risk among individuals with PHEU.
Recently, there has been increasing interest in the neurocognitive development of CHEUs. A previous meta-analysis showed that CHEUs by HIV at 12 to 24 months of age exhibited poorer expressive language and gross motor skills than CHUs. However, their cognitive development, receptive language skills, and fine motor functions were similar to those of unexposed children. This study did not establish a link between specific maternal ART and neurodevelopmental outcomes.14 Frange and colleagues (Abstract 1019) presented findings from a population-based cohort study that compared the incidence of neurodevelopmental disorders in uninfected children who were prenatally exposed to HIV and ART with unexposed children. The study analyzed data from the French National Health Data System, which included 6,667,363 live-born singletons, of whom 9035 were born to WWH and had prenatal ART exposure. The children were followed up for up to 11 years, with a mean follow-up of 5.5 years. The incidence of neurodevelopmental disorders in exposed children was higher than in the general population, even after adjusting for gestational age and sociodemographic factors, such as maternal age and employment status (HR 1.24; 95% CI, 1.04-1.47). In addition, compared with other PIs with TDF/FTC, prenatal exposure to DRV/r in combination with TDF/FTC was associated with a higher incidence of neurodevelopmental disorders and related specialized consultations. The authors highlighted the importance of careful monitoring, early detection, and management of neurodevelopmental disorders in CHEU.
Wong and colleagues (Abstract 997) investigated perinatal and infant outcomes following exposure to BIC (an alternative ART recommended for use during pregnancy in US and European treatment guidelines) in pregnancy compared with other ART. The researchers analyzed population-based, retrospective cohort study data from the Canadian Perinatal HIV Surveillance Program. BIC exposure was not associated with adverse perinatal or early infant outcomes, showing no differences in maternal HIV viral load at delivery, mode of delivery, rates of small for gestational age, perinatal transmission, or congenital anomalies.
Notes
All abstracts cited in the text appear in the CROI 2025 Abstract eBook, available online at www.CROIconference.org
The IAS-USA will identify and resolve ahead of time any possible conflicts of interest that may influence CME activities with regard to exposition or conclusion. All financial relationships with ineligible companies for the authors and planners/reviewers are below.
Financial affiliations in the past 24 months: Dr Gunaratne reported no relevant financial relationships with ineligible companies. (Updated June 2, 2025) Dr Tieu reported grant support awarded to her institution from GlaxoSmithKline. (Updated May 30, 2025) Dr Wilkin reported serving as a consultant and receiving grant support awarded to his institution from GlaxoSmithKline, ViiV Healthcare, and Merck and Co., Inc. (Updated June 2, 2025) Dr Taylor reported no relevant financial relationships with ineligible companies. (Updated June 2, 2025)
Additional References Cited in Text
JG Rosen, A Ndyanabo, H Nakawooya, et al. Incidence of health facility switching and associations with HIV viral rebound among persons on antiretroviral therapy in Uganda: a population-based study. Clin Infect Dis. 2024;78(6): 1591–1600. doi: 10.1093/cid/ciad773
PF Njau, E Katabaro, S Winters, et al. Impact of financial incentives on viral suppression among adults initiating HIV treatment in Tanzania: a hybrid effectiveness-implementation trial. Lancet HIV. 2024;11(9): e586–e597. doi: 10.1016/S2352-3018(24)00149-8
K Sikombe, A Mody, CW Goss, et al. Effect of a multicomponent, person-centred care intervention on client experience and HIV treatment outcomes in Zambia: a stepped-wedge, cluster-randomised trial. Lancet HIV. 2025;12(1): e26–e39. doi: 10.1016/S2352-3018(24)00264-9
JC Dombrowski, SR Galagan, M Ramchandani, et al. HIV care for patients with domplex needs: a controlled evaluation of a walk-in, incentivized care model. Open Forum Infect Dis. 2019;6(7): ofz294. doi: 10.1093/ofid/ofz294
KM Marks, M Kang, T Umbleja, et al. HepB-CpG vs HepB-Alum vaccine in people with HIV and prior vaccine nonresponse. The BEe-HIVe randomized clinical trial. JAMA. 2025;333(4): 295–306. doi: 10.1001/jama.2024.24490
KM Marks, M Kang, T Umbleja, et al. Immunogenicity and safety of hepatitis B virus (HBV) vaccine with a tolllike receptor 9 agonist adjuvant in HBV vaccine-naive people with human immunodeficiency virus. Clin Infect Dis. 2023;77(3): 414–418. doi: 10.1093/cid/ciad201
C Kityo, IK Mambule, J Musaazi, et al. Switch to long-acting cabotegravir and rilpivirine in virologically suppressed adults with HIV in Africa (CARES): week 48 results from a randomised, multicentre, open-label, non-inferiority trial. Lancet Infect Dis. 2024;24: 1083–1092. doi: 10.1016/S1473-3099(24)00289-5
A Avihingsanon, H Lu, CL Leong, et al. Bictegravir, emtricitabine, and tenofovir alafenamide versus dolutegravir, emtricitabine, and tenofovir disoproxil fumarate for initial treatment of HIV-1 and hepatitis B coinfection (ALLIANCE): a double-blind, multicentre, randomised controlled, phase 3 non-inferiority trial. Lancet HIV. 2023;10(10): e640–e652. doi: 10.1016/S2352-3018(23)00151-0
AM Wensing, C Charpentier, V Calvez, et al. Utilizing HIV proviral DNA to assess for the presence of HIV drug resistance. Clin Infect Dis. Published online April 3, 2025. doi: 10.1093/cid/ciaf161
H Waalewijn, AJ Szubert, RE Wasmann, et al. First pharmacokinetic data of tenofovir alafenamide fumarate and fenofovir with dolutegravir or boosted protease inhibitors in African children: a substudy of the CHAPAS-4 trial. Clin Infect Dis. 2023;77(6): 875–882. doi: 10.1093/cid/ciad267
medRXIV.org. CHAPAS-4 trial: second-line anchor drugs for children with HIV in Africa. 2025. Published April 15, 2024. Accessed May 15, 2025. https://www.medrxiv.org/content/10.1101/2024.04.12.24305333v1
CJ Wedderburn, E Weldon, C Bertran-Cobo, et al. Early neurodevelopment of HIV-exposed uninfected children in the era of antiretroviral therapy: a systematic review and meta-analysis. Lancet Child Adolesc Health. 2022;6(6): 393–408. doi: 10.1016/S2352-4642(22)00071-2