IMPLEMENTATION AND OUTCOMES OF DOLUTEGRAVIR-BASED FIRST-LINE ANTIRETROVIRAL THERAPY FOR PEOPLE WITH HIV IN SOUTH AFRICA: A RETROSPECTIVE COHORT STUDY

Background There is little data evaluating uptake of first-line dolutegravir among men and women living with HIV in low- and middle-income countries, and subsequent clinical outcomes in non-trial settings. We aimed to determine dolutegravir uptake in women, and the impact of dolutegravir on clinical outcomes in routine care in South Africa. Methods We analysed de-identified data from adults receiving first-line ART at 59 South African clinics from December 2019 to February 2022, using two distinct cohorts. In the ‘initiator cohort’, we used Poisson regression models to assess initiation with dolutegravir-based ART by gender, and associations between dolutegravir use and 12-month retention-in-care and viral suppression <50 copies/mL. In the ‘transition cohort’, consisting of adults receiving non-dolutegravir-based first-line ART in December 2019, we used Cox proportional hazards models to assess the hazard of transition to first-line dolutegravir by gender. We then used time-dependent propensity score matching to compare subsequent 12-month retention-in-care and viral suppression between people transitioned to dolutegravir, and those who had not yet transitioned at the same timepoint. Findings Of 45,392 adults initiating ART, 28,725 (63·3%) were women, median age was 31 years (IQR 26,38), and 31,264 (69·9%) initiated dolutegravir. Dolutegravir initiation was lower in non-pregnant (risk difference [RD] −18·4% [−21·6,−15·2]) and pregnant women (RD −35·4% [−42·3,−28·5]) versus men. At 12-months, retention-in-care (adjusted RD 5·2% [2·2,8·4)) and viral suppression (aRD 3·1% [1·2,5·1]) was higher in dolutegravir initiators. Among 180,956 adults receiving first-line ART in December 2019, median age was 38 years (32,45), 124,168 (68·6%) were women and 98·6% received efavirenz. By February 2022, 121,210/158,999 (76·2%) of those in care were transitioned to first-line dolutegravir. Transition was lower in women (hazard ratio 0·56, [0·56,0·57]). Among 92,318 propensity score matched people, 12-month retention-in-care (aRD 2·5% [2·1–2·9]) and viral suppression (aRD 0.8% [0.3,1.4]) were slightly higher in the dolutegravir versus non-dolutegravir group. Interpretation Women were less likely to receive dolutegravir. As dolutegravir was associated with better outcomes, the rollout should continue, with a particular emphasis on inclusion of women. Funding Wellcome Trust; Africa Oxford Initiative; International Association of Providers of AIDS Care, Bill & Melinda Gates Foundation


INTRODUCTION
Since 2018 the World Health Organization (WHO) has recommended dolutegravir-based first-line antiretroviral therapy (ART), 1 due to clinical trial evidence of increased efficacy compared to non-nucleoside reverse transcriptase inhibitors (NNRTIs). This is largely driven by better tolerability and less discontinuations, and dolutegravir is also more robust against the development of HIV drug resistance. 2,3 Initially, safety concerns regarding a potential increased risk of neural tube defects if dolutegravir was taken at conception, led the WHO to recommend restricted use among women of child-bearing potential. 1 However, new data suggesting a lower risk of neural tube defects, risk-benefit analyses from modelling studies, 4 and community advocacy 5 led WHO to remove these restrictions in July 2019. 1 By mid-2022 dolutegravir had been introduced into the preferred first-line ART regimen in approximately 108 countries. 6 To date, there are few published studies examining the impact of the above restrictions on dolutegravir uptake over time, [7][8][9] and little data comparing effectiveness of dolutegravir with alternative first-line regimens in non-trial settings in Africa. 10 In regions with high prevalence of tuberculosis co-infection, there are also important concerns regarding the use of dolutegravir with rifampicin-containing tuberculosis treatment, which reduces dolutegravir drug levels. This can be overcome by doubling the dolutegravir dose, 11 which may not always be feasible in large-scale ART programmes. Dolutegravir use is particularly pertinent in South Africa, where there are over 7·5 million PLHIV, the majority of whom are women of child-bearing potential, and tuberculosis incidence is high.
We aimed to assess the impact of the first-line dolutegravir rollout on HIV treatment outcomes in South Africa. We determined a) the extent to which women were less likely to receive dolutegravir over time, b) the relationship between dolutegravir use, and retentionin-care and viral suppression, and c) the effect of dolutegravir use on HIV treatment outcomes among people receiving concurrent tuberculosis treatment.

Study design and setting
We conducted a cohort study using de-identified, routinely collected data from 59 public sector, primary care clinics run by the eThekwini Municipality Health Unit, in the province of KwaZulu-Natal, South Africa. ART is provided free of charge in accordance with South African National Department of Health guidelines, 12 which recommended dolutegravir from December 1, 2019. Initially, ART initiations were prioritised, and women of child-bearing potential required to sign an 'acknowledgement of risk' form. On February 24, 2020, eligibility was expanded to include people already receiving first-line ART, with a riskbenefit discussion replacing the signed risk form for women of child-bearing potential. 13 In June 2021, the risk-benefit discussion for women was removed, and dolutegravir became the preferred first-line ART regimen. 14 For PLHIV with tuberculosis co-infection, dolutegravir was only recommended after completing tuberculosis treatment. Viral load testing was recommended annually, and transition to first-line dolutegravir was only recommended if people had a suppressed viral load of <50 copies/mL in the previous six months, or consecutive viral loads between 50-999 copies/mL. 12

Participants
We analysed two mutually exclusive groups; the 'initiator cohort', and the 'transition cohort'. For the initiator cohort, we evaluated dolutegravir use by gender among all PLHIV aged ≥15 years and newly initiating first-line ART (irrespective of NRTI backbone) at participating clinics, between December 1, 2019 and February 28, 2022. We excluded people with known previous ART exposure, as South African guidelines recommend reinitiating the previous regimen, which would often not be dolutegravir-based. We then analysed outcomes among those who initiated ART before December 1, 2020 and therefore had at least 12 months of follow-up time plus 90 days to assess retention-in-care before the data cut on February 28 th , 2022. We assessed viral load outcomes only among those who were retained-in-care at 12 months.
For the transition cohort, we assessed transition to first-line dolutegravir by gender among PLHIV aged ≥15 years who were in care and receiving non-dolutegravir first-line ART at participating clinics on December 1 st , 2019. We excluded those with viral load ≥1000 copies/mL in the previous 12-months, as they would not have been eligible for transition to first-line dolutegravir. We then analysed outcomes among a subset who had initiated dolutegravir before December 1, 2020, compared to people who had not initiated dolutegravir at the same timepoint, matched 1:1 using propensity score matching (appendix pp 1-2).

Data sources and data management
For patients initiating and receiving ART in the South African public sector, data on demographics, clinical status, ART and clinic visits are routinely recorded in the TIER.net 15 electronic register. TIER.net data from participating clinics was collated and de-identified by the South African National Department of Health's TB/HIV Information Systems (THIS -www.tbhivinfosys.org.za/) before being extracted for analysis.
We used demographic and clinical variables recorded at ART initiation or during follow-up.
Outcomes-We defined first-line dolutegravir use from the ART initiation regimen (initiator cohort) or the date of first use of dolutegravir in a first-line regimen (transition cohort, see appendix pp1 for full definition). We defined retention-in-care as not being recorded in TIER.net as deceased, lost-to-follow-up (defined as 90 days late for a visit by the South African ART programme, with date of last visit used as date of loss-to-follow-up) or 'transferred out' to another clinic (as we could not access or link to data at other clinics to establish retention-in-care). We used viral suppression <50 copies/ml, as per clinical trials [16][17][18] and South African guidelines. In this programmatic setting, not all patients had annual viral loads. Therefore, in both cohorts we included viral loads five to 18 months after baseline, and used the result closest to 12 months.

Statistical analysis
Among the initiator cohort, we first used univariable Poisson regression models with robust standard errors to estimate relative risks for the association between gender, and the outcome of dolutegravir initiation. We also conducted sensitivity analyses, defined a priori, to assess whether the effect of gender on dolutegravir use was modified by age at ART initiation, and by time. We then used univariable and multivariable Poisson regression models with robust standard errors to assess the association between being initiated on dolutegravir, and 12-month retention-in-care, and viral suppression. To assess whether tuberculosis treatment impacted on viral suppression with dolutegravir, we did a sensitivity analyses, defined a priori, with an interaction term between dolutegravir use and tuberculosis status.
Among the transition cohort, we used univariable Cox proportional hazards models to assess the association between gender and time to transition to a dolutegravir-based first-line regimen. We started follow-up time from December 1, 2019, and censored people at date of loss-to-follow-up, death, transfer to another clinic, or switch to second-line ART. We also censored people with a viral load of >1000 copies/mL during follow-up, as they would not have been eligible for immediate transition to first-line dolutegravir.
Comparing outcomes between people transitioned and not transitioned to dolutegravir is complicated by the lack of a clear baseline time in those whose treatment remains unchanged, and potential underlying differences between the two groups. We addressed these issues by emulating a 'target trial', in which each person transitioned to dolutegravir was matched to a control who had not yet been transitioned at the same timepoint. 19 To increase comparability between the two groups, we matched participants 1:1 using a time-dependent propensity score of the log-hazard of transition to dolutegravir, 20,21 with direct matching by time since most recent viral load, and whether the participant was in a differentiated ART delivery programme. Individuals could only be matched once, and we only matched participants transitioned to dolutegravir before December 1 st , 2020, to allow 12 months of follow-up time. Further details are provided in the appendix pp 1-2. We then used multivariable Poisson regression models to compare the outcomes of 12-month retention-in-care and viral suppression between people transitioned to dolutegravir, and their matched controls.
In both the initiation and transition cohort, the primary viral load analysis was an intentionto-treat analysis, with a secondary 'as-treated' analysis that excluded people who changed ART regimen after baseline. Dolutegravir rollout and clinical outcomes may have varied by clinic, and we accounted for this using robust standard errors. We did not adjust for multiple comparisons. We present both risk ratios and risk differences to demonstrate both the relative strength of an association, and the absolute difference. We included a specific category for missing baseline CD4 count data. We analysed data using R 4.2.0 (R Foundation for Statistical Computing, Vienna, Austria) and STATA 14.0 (StataCorp, Texas, USA).

Ethical approval
This work was approved by the University of Kwazulu-Natal Biomedical Research Ethics Committee (BE646/17), the KwaZulu-Natal Provincial Health Research Ethics Committee (KZ_201807_021), the THIS Data Request Committee, and the eThekwini Municipality Health Unit, with a waiver for informed consent for analysis of de-identified, routinely collected data.

Role of the funding source
The funders had no role in the study design, data collection, analysis, interpretation, writing of the paper, or the decision to submit for publication.

DISCUSSION
In this large cohort study from 59 South African clinics, women were less likely to receive dolutegravir compared to men, with the effect strongest early in the rollout and among younger women. In both people initiating and already receiving first-line ART, dolutegravir use was associated with better 12-month retention-in-care and viral suppression. The association between dolutegravir use and viral suppression was stronger among people receiving concurrent tuberculosis treatment when initiating ART, and among people transitioned to dolutegravir with most recent viral load at baseline ≥200 copies/mL. We provide longer term data compared to studies from earlier in the dolutegravir rollout, which showed that initially younger women were less likely to receive dolutegravir. 7-9 A multi-site study of 134,672 PLHIV from 11 countries up to March 2020 (eight months after the WHO recommended dolutegravir for all PLHIV), found that in people aged 16-49, dolutegravir use was lower among women compared to men, but longer term trends were not assessed. 7 In our study, younger women were less likely to receive dolutegravir up to two years after WHO recommended dolutegravir for all. This difference decreased over time and disappeared shortly after South African guidelines recommended dolutegravir for all in June 2021, with the hazard of transition to dolutegravir becoming higher among women compared to men in the subsequent months; a likely 'catch-up' effect. However, by the end of the follow-up period, women remained overall less likely to be on dolutegravir than men, and 15% of people remained on non-dolutegravir first-line ART.
Retention-in-care at 12-months was low at 63·6% among people newly initiating ART in our cohort, with a higher proportion (14·2%) transferring to other clinics than seen in a previous study (6·1%), 22 which may reflect mobility due to COVID-19. Our finding that first-line dolutegravir was associated with better retention-in-care may be due to increased tolerability, which could be particularly important early in treatment. This is similar to findings from clinical trials, in which the superior efficacy of dolutegravir has largely been driven by reduced discontinuations from adverse events. [16][17][18] However, better retention was partly driven by more clinic transfers among people receiving efavirenz, reasons for which are not clear, but could represent people seeking more tolerable ART and better treatment at another clinic. We also found better 12-month viral suppression with dolutegravir, particularly in people initiating ART, with the risk difference within the range seen in two clinical trials in Africa. 16,18 Among those transitioning from other first-line ART regimens (a group in which no clinical trials have been conducted), we only found improved viral suppression with dolutegravir among those with most recent viral loads ≥200 copies/mL. This could be due to HIV drug resistance against efavirenz among people with low level viraemia, better efficacy of dolutegravir at lower levels of adherence, or improved adherence due to better tolerability of dolutegravir.
There is little data from non-trial settings directly comparing outcomes between dolutegravir and other regimens, particularly in Africa. Public health data from Brazil showed good safety outcomes, 23 and better 12-month viral suppression among people initiated on dolutegravir compared to efavirenz, 24 but the proportion on dolutegravir was low (10·5%), and there was no assessment of retention-in-care. Studies from Malawi, Lesotho, and Uganda suggest low levels of dolutegravir HIV drug resistance mutations, and high levels of viral suppression, among people transitioned to dolutegravir, but do not compare results with people remaining on non-dolutegravir regimens. [25][26][27] A retrospective cohort study of 3,108 people from four African countries found that people transitioned to dolutegravir had better viral suppression compared to those who remained on the same first or second-line regimen. 10 However, analyses such as this are susceptible to bias because it is difficult to choose an appropriate baseline time zero in the control group whose treatment remains unchanged, resulting in baseline timepoints and viral load schedules differing between the two groups, and potential for immortal time bias. We emulated a target trial to overcome this, and present the largest analysis comparing outcomes after transition to dolutegravir among people on first-line ART, who are the largest group of people who will use dolutegravir globally. South African guidelines recommend efavirenz rather than dolutegravir among people initiating ART who are receiving rifampicin containing tuberculosis treatment. However, we found that over 50% (n=1305) did receive dolutegravir, and among this group, the beneficial effect of dolutegravir on viral suppression was even stronger. The INSPIRING trial demonstrated good tolerability and acceptable viral suppression among people receiving tuberculosis treatment who were given double dose dolutegravir. 11 While the extent of dolutegravir double dosing is not recorded in our data, our findings provide re-assurance that co-treating TB-HIV co-infection did not compromise HIV outcomes in a high TB burden programmatic setting. Our findings are supported by a smaller cohort study including 465 people receiving dolutegravir co-treatment alongside tuberculosis treatment, which found better viral suppression compared to people receiving co-treatment with efavirenz. 28 Our study is the largest to evaluate dolutegravir uptake and compare subsequent treatment outcomes against non-dolutegravir based regimens in a public health programme. We used data from routine public sector clinics, which provide care according to South African Department of Health guidelines, and used programmatic outcome definitions, 12 making our findings more generalisable to other public sector settings (although our clinics were limited to one urban district). We directly compared both retention-in-care and viral suppression between dolutegravir and non-dolutegravir regimens, with precise estimates due to the large sample size. Our use of an emulated target trial in the transition cohort, with propensity score matching to balance the dolutegravir and non-dolutegravir groups, should increase comparability, although we cannot rule out residual unmeasured confounding. We are likely to have overestimated loss-to-follow-up as mortality and 'silent transfers' to other clinics are underestimated in TIER.net. 29 As we used routinely collected data, we were unable to search for silent transfers to other clinics, and did not have consent to search for deaths on the national registry. We assessed 12-month treatment outcomes, and further work will be required to assess if outcomes remain similar after longer follow-up. We were unable to assess the dose of dolutegravir used in people with tuberculosis, HIV drug resistance and adverse events such as weight gain, as these are not recorded in TIER.net.
Our findings are important as they demonstrate the extent to which women were excluded from the early dolutegravir rollout. They also provide reassurance that in programmatic settings, dolutegravir is associated with similar or better outcomes than efavirenz, reflecting findings from clinical trials. While improvements in retention-in-care and viral suppression with dolutegravir were modest, incremental gains are important in reaching the 95-95-95 targets. However, overall loss-to-follow-up of 20% by 12 months among people newly initiating ART demonstrates that early retention-in-care remains a key challenge for HIV programmes, and this could limit the potential benefit of improved ART regimens. Strategies to identify and support people at risk of loss to follow up are therefore needed. Our findings support ongoing efforts to continue the transition to dolutegravir, and to remove restrictions on dolutegravir use among people being treated for tuberculosis. Efforts should particularly focus on ensuring that women receive updated safety information and are provided the opportunity to use dolutegravir without restrictions. More broadly, strategies to introduce newer antiretrovirals at scale should ensure that the necessary safety evidence is been generated as quickly as possible prior to rollout, and that pregnant women are included in drug trials where possible. Further research is needed in non-trial settings to assess reasons for not transitioning to dolutegravir, 30 adverse events such as weight gain and metabolic consequences (which seem to disproportionately affect women), 31 the impact of transition to dolutegravir among people with viraemia ≥1000 copies/mL, 32 and the use of dolutegravir in second-line regimens. 33 In conclusion, we found that young women were less likely to receive dolutegravir until September 2021, and that people receiving dolutegravir had better retention-in-care and viral suppression compared to efavirenz. Efforts to transition to dolutegravir should continue.

Evidence before this study
We searched PubMed with no language restrictions on October 5, 2022, with the terms [dolutegravir AND (rollout OR implementation OR outcomes) AND Africa] and identified additional studies using hand searches of reference lists and citing papers.
After concerns were raised in 2018 regarding a potential association between dolutegravir use at conception and neural tube defects, there have been three observational studies comparing the uptake of dolutegravir between women and men in Africa. Two studies early in the rollout from found that approximately 60% of men compared to 30% of women had been transitioned to dolutegravir, while a smaller more recent study found that by September 2021, 68% of women had been transitioned to dolutegravir, compared to 80% of men.
Regarding clinical outcomes on dolutegravir, a large network meta-analysis of 68 randomised controlled trials (four of which directly compared dolutegravir 50mg daily with efavirenz among people newly initiating first-line ART) found that dolutegravir had better viral suppression, less HIV drug resistance and fewer discontinuations compared to first-line efavirenz. There are no trials assessing transition to dolutegravir among people already receiving efavirenz.
Despite the large scale dolutegravir rollout in low-and middle-income countries, there have been few assessments of the effectiveness of dolutegravir compared to other firstline regimens in non-trial settings. Among people initiating first-line ART in Brazil, 12-month viral suppression <50 copies/mL was higher among people initiated on dolutegravir versus efavirenz, but people with no 12-month viral load were excluded and loss-to-follow-up was not assessed. A retrospective cohort study among 3,108 people in four African countries found higher hazard of viraemia >1000 copies/ among people who remained on non-dolutegravir regimens compared to those transitioned to dolutegravir. However, this analysis is susceptible to multiple biases, including systematic differences in viral load schedules and the start of follow-up time between the two groups, and the use of events after baseline to determine inclusion in the time-to-event analysis.
Regarding use of dolutegravir among people with tuberculosis, rifampicin has been found to reduce dolutegravir levels. This may be overcome by doubling the dose of dolutegravir to 50mg twice daily, an approach that was found to be well tolerated in the INSPIRING trial. In non-trial settings, a cohort study among 3563 people with HIV receiving rifampicin-containing tuberculosis treatment found that those on dolutegravirbased regimens had better viral suppression compared to those who received efavirenz.

Added value of this study
We present the largest study to date to assess dolutegravir uptake in women versus men, and to compare outcomes with non-dolutegravir first-line regimens in routine care settings. We show that younger women were less likely to be initiated or transitioned to dolutegravir early in the rollout, but this difference resolved once South African guidelines recommended dolutegravir for all people living with HIV (PLHIV), irrespective of child-bearing potential. Among both people initiating first-line ART, or already receiving first-line ART, dolutegravir use was associated with better 12-month retention and viral suppression. The association between dolutegravir and better viral suppression was even higher among people being treated for tuberculosis.

Implications of all the available evidence
Taken together, these findings support the ongoing rollout of dolutegravir for first-line ART in low-and middle-income countries, including among people being treated for tuberculosis. Our findings also suggest that, due to initial safety concerns, women have been excluded from receiving an effective ART regimen. Therefore, particular efforts should be made to ensure that women who previously did not receive dolutegravir, are now offered dolutegravir alongside updated evidence regarding safety and effectiveness. Further evidence from routine care settings regarding adverse events, in particular weight gain, are needed.