Children and adolescents living with HIV (CALHIV) continue to have lower viral suppression rates than adults.^1,2 In South Africa, there are approximately 150 000 CALHIV and the viral suppression rate in those with available data is approximately 50%.^3,4 Children aged 1–4 years old have the lowest rate of viral suppression, at 36.3%.⁴ Delayed access to appropriate drugs and fixed-dose combinations (FDC) for children and adolescents across all age groups contribute to this poor suppression rate.⁵

Dolutegravir, a preferred integrase strand transfer inhibitor with non-inferior or superior efficacy when compared to efavirenz, raltegravir, and boosted protease inhibitor-based regimens, was approved for use in adults in 2013.^6,7 However, paediatric-appropriate formulations of dolutegravir were only licensed by the Food and Drug Administration in 2020 based on data from the IMPAACT P1093 and ODYSSEY studies.^{8,9,10,11,12,13,14,15} The ODYSSEY and CHAPAS-4 studies, along with routine data, confirm the efficacy of dolutegravir in CALHIV with persistent viral non-suppression following a switch to dolutegravir.^12,16,17

In line with the 2018 WHO guidelines, the 2019 South African guidelines recommended dolutegravir as the preferred initial regimen, as well as for switching in individuals with persistent viral non-suppression.^18,19,20,21 However, the 10 mg scored dispersible formulation, appropriate for use in children weighing < 20 kg, only became available in 2022.²²

Switching to dolutegravir in the presence of persistent viral replication, as well as switching during childhood or adolescence, are risk factors for developing dolutegravir resistance.²³

In adults with persistent viral replication switching to dolutegravir, maintaining a backbone of tenofovir disoproxil fumarate (TDF) with lamivudine or emtricitabine is superior to switching to zidovudine with lamivudine,^{24,25,26,27,28} however, adult doses of TDF can only be used in children aged ≥ 10 years and weighing ≥ 30 kg.²¹ In the CHAPAS-4 study, children with persistent viral replication randomised to tenofovir alafenamide fumarate (TAF) and lamivudine with dolutegravir were more likely to achieve viral suppression than those receiving abacavir or zidovudine and lamivudine with dolutegravir.¹⁶ Access to TAF remains limited and dosing has not been established across all weight bands.

In South African CALHIV with persistent viral non-suppression weighing < 30 kg, abacavir is recycled.²¹ It remains unclear whether abacavir recycling has the same efficacy as TDF or TAF when switching to dolutegravir in second-line therapy in the programmatic setting.

We aimed to study the outcomes of CALHIV attending Tygerberg Hospital in South Africa who were switched to dolutegravir despite persistent viral non-suppression, and to identify risk factors associated with further persistent non-suppression in this vulnerable cohort.

We performed a retrospective folder review of CALHIV aged < 18 years managed at Tygerberg Hospital Paediatric Infectious Diseases Clinic. Eligible participants were CALHIV with persistent viral non-suppression who switched to a dolutegravir-based regimen between 01 January 2021 and 30 June 2024. Persistent viral non-suppression was defined as at least one episode of two or more viral loads ≥ 1000 copies/mL measured at least 3 months apart, and a viral load of > 50 copies/mL at the time of switch to dolutegravir. Children who interrupted therapy prior to the switch to dolutegravir were also included.

As hospitalised CALHIV who initiated dolutegravir as inpatients may have been referred to antiretroviral therapy (ART) facilities closest to their home, we included only those with at least one follow-up visit after switching to dolutegravir.

Children were identified using the clinic database. Clinical data and prior ART history were recorded from paper-based and electronic medical records. Laboratory data including CD4+ T-lymphocyte cell counts and percentages, viral loads and results of drug resistance tests were obtained from the National Health Laboratory Service.

We collected data retrospective data up to 31 October 2024. Data were originally collected up until 31 August 2024; however, because of a national cybersecurity incident in 2024 that disrupted routine follow-up testing, viral load data collection was extended by 2 months to capture delayed or missed measurements, with approval from the Health Research Ethics Committee.

Baseline viral load was defined as the viral load measurement closest to the date of switching to dolutegravir, within the preceding 6 months. Children without this result were not assigned a baseline viral load value. Baseline CD4+ T-lymphocyte count and percentage were defined as the measurement closest to switching, within 6 months before or 1 month after the switch. Viral suppression in children on dolutegravir was defined as a viral load ≤ 50 copies/mL at the last recorded measurement before 31 October 2024, at the date of transfer out, or at the date of last clinical encounter in cases of treatment interruption. Where possible, we categorised the interruption interval as less than 1 month, 1–3 months, and more than 3 months. Time on dolutegravir was calculated from the date of switch to dolutegravir to 31 October 2024, transfer out, or the date of last clinical encounter in cases of treatment interruption.

Weight-for-age, height-for-age, and body mass index (BMI)-for-age Z-scores were reported at the time of switch and at the last clinic visit up to 31 August 2024. The Z-scores were calculated using the WHO child growth standards and AnthroCalc version 3.0.2 mobile application for children aged less than 5 years and AnthroPlus for older children.^29,30

Cumulative resistance profiles were analysed using the Stanford HIV Drug Resistance Database and composite resistance scores were reported.³¹

Data were analysed using Statistical Package for Social Sciences (SPSS) version 29 (IBM Corp., Armonk, New York, United States). For continuous variables, we report measures of central tendency (mean, median, mode) and dispersion (standard deviation, range and interquartile range [IQR]) based on the distribution of the data. Categorical variables were summarised using frequencies and percentages. Univariate analyses were performed using Chi-Square tests and univariate binary logistic regression to screen risk factors that might be related to our outcome variable. Variables identified in univariate analyses were entered into a multivariable binary logistic regression model to identify independent risk factors for our outcome variable. The results are reported as odds ratios (OR) with 95% confidence intervals (CI) to quantify the strength of the associations.

At the time of screening, 218 patients attended the clinic. Of these, 39 children were eligible for the study, of whom 18 (46.2%) were girls. The median age at ART initiation was 1 year (interquartile range [IQR] 0.18; 1.83) (Table 1). Children were switched to a dolutegravir-based regimen at a median age of 7.2 years (IQR 3.02; 9.73). The rate of viral suppression to ≤ 50 copies/ml was 55.3% (Table 2).

Notably, eight children (20.5%) were hospitalised at the time of the switch, and six children (15.4%) were on rifampicin-based anti-tuberculosis therapy (Table 1).

Prior to switching to dolutegravir, all children were receiving abacavir and lamivudine. Nearly all of the children (n = 38, 97.4%) had received one or more protease inhibitors (Table 1). Twelve children (30.8%) had never achieved a documented viral load of ≤ 1000 copies/mL prior to initiating dolutegravir (Table 1). Interruptions in care were common, with 23 children (59.0%) experiencing at least one interruption in treatment exceeding 1 month, prior to the switch (Table 1).

Only six children (15.4%) were switched to TDF/lamivudine/dolutegravir, and no children received zidovudine (Table 2). Among the 21 children (55.3%) who achieved viral suppression by the end of the study, eight (38.1%) had been on therapy for < 6 months (Table 2). Of the 17 children (44.7%) with persistent viral non-suppression (Table 2), seven (41.2%) had a viral load between 50 and < 400 copies/mL. Six of the seven children (85.7%) on treatment ≤ 6 months had low-level viraemia (50–400 copies/mL).

Abacavir was recycled in 31 children (available viral load for 30/31), of whom 50.0% achieved viral suppression (Table 2 and Table 3). The difference in suppression rates between children receiving a TDF-containing backbone and an abacavir-containing backbone was not statistically significant. However, the number of children receiving a TDF-containing backbone was small, limiting meaningful comparison. Duration of treatment prior to switch, age at treatment initiation, age at switch to dolutegravir, and availability of resistance testing before switch were not predictive of persistent viral non-suppression (Table 3). Fifteen children (38.5%) experienced documented interruptions in care after switching to dolutegravir. Of these, 12 (80.0%) had also interrupted therapy at least once prior to the switch.

Twelve children (30.8%) had one or more resistance tests (Table 1). Only one child had resistance testing performed after switching to dolutegravir (Table 4). Of the eight children receiving dolutegravir with an abacavir-containing backbone who had a resistance test result available, five (66.7%) had some documented resistance to abacavir. Four of those five children suppressed.

After switching to a dolutegravir-based regimen, eight children (21.5%) required hospitalisation (Table 2). We did not review the indications for hospitalisation. Two children (5.1%) developed tuberculosis, including one case of rifampicin-resistant disease (Table 2). Among children with available data, the median BMI Z-score at the last documented clinic visit was 0.37 (IQR –0.29; 1.17), representing an increase from a median BMI Z-score of –0.04 (IQR –0.71; 0.96) at the time of dolutegravir initiation (Figure 1 and Table 2). Dolutegravir was not discontinued in any child because of an adverse event.

We present South African programmatic data on the virological outcomes of children switched to a dolutegravir-based regimen while experiencing persistent viral non-suppression. At median follow-up of 11.03 months (IQR 3.30; 22.30), 55.3% of children achieved viral suppression to ≤ 50 copies/mL. However, a significant number of children had only a single follow-up viral load measurement, often obtained less than 6 months after switching to dolutegravir.

It appears that the outcomes in our cohort are less favourable than those reported in the ODYSSEY and CHAPAS-4 trials.^12,16 In CHAPAS-4, at 48 weeks, more than 90% of children achieved viral suppression to < 400 copies/mL (primary outcome). The study also showed that more than 80% of children achieved suppression to < 60 copies/mL from approximately 6 weeks after initiation of dolutegravir, which was sustained throughout the main study period and extended follow-up.¹⁶ Similarly, in ODYSSEY Part B, among the 196 children switched to dolutegravir while experiencing persistent viral non-suppression, 81% (95% CI 75–86) achieved viral loads < 50 copies/mL and 89% (95% CI 8 4–93) achieved viral loads < 400 copies/mL at 96 weeks.¹² This highlights important differences between clinical trial populations and real-world programmatic cohorts. Our cohort consisted of younger children compared with those enrolled in CHAPAS-4 and ODYSSEY Part B.^12,16 At dolutegravir initiation, the median age in our cohort was 7.19 years (IQR 3.02; 9.73), with 15 children (38.5%) aged < 5 years and only nine (23.1%) aged ≥ 10 years. In contrast, CHAPAS-4 reported a median age of 11 years (IQR 8; 13), with only six children (2.6%) aged 3–4 years and the majority (139, 60.7%) aged 10–15 years. Similarly, in ODYSSEY (Parts A and B) the median age was 12.2 years (IQR 9.2; 15.1). These age differences were also reflected in weight band distributions. The median weight in CHAPAS-4 was 27 kg (IQR 21.3; 34.0), and in ODYSSEY (Parts A and B), 30.4 kg (IQR 23.7; 43.7).^12,16 In contrast, 20 children (51.28%) in our cohort weighed < 20 kg, while only nine children (25.0%) weighed ≥ 30 kg. This limited available therapeutic options, as lower weights precluded the use of the once-daily FDC tablets available in the South African HIV programme at the time. The recent introduction of a paediatric dispersible abacavir/lamivudine/dolutegravir FDC (pALD), together with the availability of generic dolutegravir-based FDCs with a TAF-containing backbone, is expected to improve this situation. Although pALD is dispersible, children using pALD still need to take three to six tablets per day.³²

Although our cohort was relatively young, the number of children was insufficient to draw firm conclusions regarding outcomes in those weighing < 14 kg. In ODYSSEY Part B, 13 children weighed < 14 kg, of whom seven were assigned to the dolutegravir arm.¹⁴ Across ODYSSEY Parts A and B, treatment failure at 96 weeks occurred in 31% of children weighing < 14 kg on dolutegravir, compared with 13.4% among those weighing ≥ 14 kg.^12,14 This is consistent with existing evidence that infants and younger children take longer to achieve viral suppression and are possibly at higher risk of persistent viral non-suppression.

In ODYSSEY Parts A and B, there is no statistically significant difference in weight gain between dolutegravir and standard-of-care groups at follow-up. The mean differences in weight (dolutegravir minus standard of care) were 1.0 kg (95% CI –0.2 to 2.2; P = 0.0950) in the ≥ 14 kg cohort at 240 weeks’ follow up, and –0.1 kg (95% CI –0.8 to 0.7; P = 0.83) in the < 14 kg cohort at 192 weeks’ follow up.³³ We present short-term weight outcomes, with the median weight-for-age Z-scores increasing from –1.145 (–1.84; –0.11) at switch to –0.56 (–1.48; 0.06), suggesting a return to health.

Although resistance to dolutegravir is rare in children initiating dolutegravir-based first-line therapy, it occurs more frequently among those who switch to dolutegravir in the context of viral failure.²³ In both CHAPAS-4 and ODYSSEY Part B, treatment failure was not primarily driven by dolutegravir resistance.^12,16 In CHAPAS-4, two of nine (22%) children receiving dolutegravir who experienced persistent viral non-suppression and underwent resistance testing had evidence of dolutegravir resistance. Notably, both had received zidovudine as part of their ART regimen.³⁴ Similarly low rates of resistance were reported in ODYSSEY Part B, where 18% (4/22) of children with persistent viral non-suppression demonstrated resistance to integrase strand transfer inhibitors.¹² We were unable to assess the development of dolutegravir resistance in our cohort, as only one child underwent resistance testing following the switch to dolutegravir. This child had the M184V mutation prior to switch but, after 28 months of treatment with TDF, lamivudine and dolutegravir, repeat resistance testing did not demonstrate integrase resistance or a detectable M184V mutation, suggesting non-adherence. This was the only child receiving TDF that experienced persistent viral non-suppression. A key distinction between our cohort and those enrolled in clinical trials is prior ART exposure. Most children in our cohort were protease inhibitor experienced, whereas participants in CHAPAS-4 and ODYSSEY Part B largely had ongoing viraemia while receiving non-nucleoside reverse transcriptase inhibitor-based regimens.^16,35 While this difference is unlikely to influence the outcomes of dolutegravir-based regimens, it may have implications for further management if protease inhibitor resistance went undetected.

In CHAPAS-4, children receiving TAF achieved higher rates of viral suppression than those receiving abacavir or zidovudine,¹⁶ findings consistent with adult data supporting the reuse of TDF over zidovudine. However, TAF was not available for paediatric use locally, and very few children in our cohort met weight thresholds for TDF. Although the proportion of children achieving viral suppression on TDF (5/6) was higher than among those receiving abacavir (15/30), the numbers are small and this difference did not reach statistical significance. However, in light of the CHAPAS-4 study results, the urgent need to improve access to TAF to optimise treatment outcomes for children switching to new ART regimens in order has become evident.

Viral suppression in this real-world cohort was lower than that reported in clinical trials. However, given the small sample size and observational study design, generalisabilty of the data is limited. That viral suppression remained suboptimal despite switching to simpler and more tolerable regimens suggests that regimen optimisation must be supported by social interventions to further support adherence.

Our findings highlight the potential importance of more comprehensive models of care, including community-based, financial and psychosocial support to better support CAHIV and their families.