Cost-effectiveness of broadly neutralizing antibody prophylaxis for HIV-exposed infants in sub-Saharan African settings

Introduction: Infant HIV prophylaxis with broadly neutralizing anti-HIV antibodies (bNAbs) could provide long-acting protection against vertical transmission. We sought to estimate the potential clinical impact and cost-effectiveness of hypothetical bNAb prophylaxis programmes for children known to be HIV exposed at birth in three sub-Saharan African settings. Methods: We conducted a cost-effectiveness analysis using the CEPAC-Pediatric model, simulating cohorts of infants from birth through death in Côte d’Ivoire, South Africa and Zimbabwe. These settings were selected to reflect a broad range of HIV care cascade characteristics, antenatal HIV prevalence and budgetary constraints. We modelled strategies targeting bNAbs to only WHO-designated “high-risk” HIV-exposed infants (HR-HIVE) or to all HIV-exposed infants (HIVE). We compared four prophylaxis approaches within each target population: standard of care oral antiretroviral prophylaxis (SOC), and SOC plus bNAbs at birth (1-dose), at birth and 3 months (2-doses), or every 3 months throughout breastfeeding (Extended). Base-case model inputs included bNAb efficacy (60%/dose), effect duration (3 months/dose) and costs ($60/dose), based on published literature. Outcomes included paediatric HIV incidence and incremental cost-effectiveness ratios (ICERs) calculated from discounted life expectancy and lifetime HIV-related costs. Results: The model projects that bNAbs would reduce absolute infant HIV incidence by 0.3–2.2% (9.6–34.9% relative reduction), varying by country, prophylaxis approach and target population. In all three settings, HR-HIVE–1-dose would be cost-saving compared to SOC. Using a 50% GDP per capita ICER threshold, HIVE-Extended would be cost-effective in all three settings with ICERs of $497/YLS in Côte d’Ivoire, $464/YLS in South Africa and $455/YLS in Zimbabwe. In all three settings, bNAb strategies would remain cost-effective at costs up to $200/dose if efficacy is ≥30%. If the bNAb effect duration were reduced to 1 month, the cost-effective strategy would become HR-HIVE–1-dose in Côte d’Ivoire and Zimbabwe and HR-HIVE–2-doses in South Africa. Findings regarding the cost-effectiveness of bNAb implementation strategies remained robust in sensitivity analyses regarding breastfeeding duration, maternal engagement in postpartum care, early infant diagnosis uptake and antiretroviral treatment costs. Conclusions: At current efficacy and cost estimates, bNAb prophylaxis for HIV-exposed children in sub-Saharan African settings would be a cost-effective intervention to reduce vertical HIV transmission. We used the validated Cost‐Effectiveness of Preventing AIDS Complications–Pediatric (CEPAC‐P) microsimulation model to project the clinical and economic impacts of hypothetical national bNAb HIV prophylaxis programmes for children known to be HIV exposed at birth in Côte d’Ivoire, South Africa and Zimbabwe [14, 15, 16]. We selected these countries to reflect a broad range of HIV epidemic characteristics with varying HIV prevalence, ART coverage, breastfeeding duration, early infant diagnosis (EID) uptake, healthcare costs and cost constraints (Table A5). Simulated infants enter the model at birth and face an initial perinatal risk and monthly postnatal risks of acquiring HIV. These risks are based on maternal HIV status, ART use, HIV viral load and breastfeeding practices. Children who acquire HIV draw a CD4% from a user‐defined distribution; without effective treatment, CD4% declines, which increases monthly risks of opportunistic infections (OIs) and AIDS‐related death. All known HIV‐exposed infants encounter opportunities for EID at multiple time points consistent with country‐specific guidelines and can undergo HIV testing after the development of an OI. Modelled children with HIV start ART immediately upon diagnosis and linkage to care; successful treatment leads to CD4% gains and reduced OI and mortality risks (see online Appendix). We followed Consolidated Health Economic Evaluation Reporting Standards in line with best‐practice advisories for cost‐effectiveness analyses (Table A4) [17]. This study was approved by the Mass General Brigham Human Research Committee. We added an infant prophylaxis module to the CEPAC‐P model (Figure A1). The impact of prophylaxis on intrapartum transmission is modelled as a multiplier on perinatal transmission risks. For breastfeeding children, prophylaxis eligibility is evaluated each month, based on age and maternal characteristics; eligible children experience a probability of access and adherence to prophylaxis. If a prophylaxis dose is received, an efficacy multiplier reduces postnatal transmission risks for a user‐specified number of months. To reflect outcomes for all known HIV‐exposed children, we modelled two sub‐cohorts in each setting: known HIV‐exposed infants who are “high‐risk” at birth by World Health Organization (WHO) criteria (e.g. infants whose mothers experienced incident HIV infection during pregnancy, received less than 4 weeks of ART before delivery or had an HIV viral load >1000 c/ml near delivery) and known HIV‐exposed infants who are “non‐high‐risk” (Figure 1) [18]. We assumed all infants known to be HIV exposed at birth who are truly high‐risk have been identified as such and varied this assumption in sensitivity analyses. We also did not assume additional costs for identifying high‐risk infants. Outcomes for each sub‐cohort were aggregated and weighted based on sub‐cohort size to produce outcomes for the overall cohort of all children known to be HIV exposed at birth, including children who acquire HIV (irrespective of diagnosed infection) (see online Appendix). To examine the impact of hypothetical bNAb prophylaxis programmes focused on different target populations, we modelled strategies in which bNAbs are offered only to the sub‐cohort of high‐risk HIV‐exposed infants (HR‐HIVE strategies) and strategies in which bNAbs are offered to both modelled sub‐cohorts, comprising the entire population of all known HIV‐exposed infants (HIVE strategies; Figure 1). All known HIV‐exposed children are eligible for standard of care (SOC) oral antiretroviral prophylaxis, that is 6 weeks of nevirapine for non‐high‐risk infants and 12 weeks of dual nevirapine and zidovudine for high‐risk infants [18]. Within each target population, we investigated the impact of varied bNAb dosing approaches, including zero doses (SOC), one dose (at birth), two doses (at birth and 3 months) and extended bNAb dosing every 3 months throughout breastfeeding. Hybrid dosing approaches, in which high‐risk and non‐high‐risk infants are eligible to receive a different number of bNAb doses, were also explored (see online Appendix; Figure A2). In the model, we do not require HIV testing prior to bNAb administration; however, prophylaxis is stopped if the infant tests positive through routine HIV testing. We modelled two separate cohorts of infants known to be HIV exposed at birth: high‐risk infants (by WHO criteria) and non‐high‐risk infants [18]. Outcomes from these two modelled cohorts were weighted by prevalence of high‐risk characteristics to generate overall clinical and economic outcomes for a population of all infants with known HIV exposure in each country setting evaluated (see online Appendix). Considering outcomes for the entire population of infants with known HIV exposure, we evaluated seven bNAb implementation strategies, which reflected assumptions regarding both the target population for bNAb administration (i.e. only high‐risk infants [HR‐HIVE] or all known HIV‐exposed infants [HIVE]) and the bNAb dosing approach (i.e. standard of care [SOC] oral infant prophylaxis without bNAbs, one dose of bNAbs at birth, two doses of bNAbs at birth and 3 months, and extended dosing of bNAbs throughout breastfeeding). In all strategies examined, all known HIV‐exposed infants were assumed to be eligible for SOC oral infant prophylaxis [18]. Abbreviations: ART, antiretroviral therapy; bNAb, broadly neutralizing antibody; SOC, standard of care; VL, viral load; WHO, World Health Organization. †All modelled outcomes are presented for the entire population of all infants with known HIV exposure, reflecting a weighted average of outcomes between the high‐risk and non‐high‐risk modelled cohorts. ‡The bNAb effect duration was altered during sensitivity analysis and the frequency of administration was adjusted accordingly. Modelled outcomes include cumulative HIV incidence and 5‐year survival, as well as discounted (3%/year) and undiscounted life expectancy and lifetime per‐person HIV‐related costs (in 2019 US dollars [USD]) from the healthcare system perspective. We chose a lifetime time horizon to capture the full clinical and economic impacts of averting HIV infection among children. We also multiplied projected cumulative HIV incidence by the HIV‐exposed birth cohort size to estimate the number of HIV infections that would occur and would be averted by each strategy relative to SOC. To calculate incremental cost‐effectiveness ratios (ICERs) in $/year‐of‐life saved (YLS), we ordered strategies by ascending life expectancy and divided the difference in discounted costs by the difference in discounted life expectancy between non‐dominated strategies. Given controversies regarding cost‐effectiveness thresholds in resource‐limited settings, we identified the most effective strategy with an ICER below thresholds of both 20% and 50% of gross domestic product (GDP) per capita (see online Appendix; Table A3). Country‐specific HIV prevalence and maternal and infant HIV care cascade characteristics were informed by UNAIDS estimates, Demographic and Health Surveys, Multiple Indicator Cluster Surveys and published literature (Table 1 and Table A5) [1, 6, 7, 8, 10, 12, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46, 47, 48, 49, 50, 51, 52, 53, 54, 55, 56, 57, 58, 59, 60, 61, 62, 63, 64]. The three settings evaluated in this analysis were chosen to reflect a range of maternal awareness of chronic HIV infection (Côte d’Ivoire: 92%, South Africa: 99% and Zimbabwe: 98%) [1, 26, 35, 36, 53], ART uptake in pregnancy (Côte d’Ivoire: 87%, South Africa: 98% and Zimbabwe: 93%) [1], mean breastfeeding duration (Côte d’Ivoire: 14 months, South Africa: 6 months and Zimbabwe: 13 months) [35, 36, 51, 56] and 2019 GDP per capita (Côte d’Ivoire: $2276, South Africa: $6001 and Zimbabwe: $1464) [65]. The proportion of infants within each sub‐cohort was calculated using estimates of maternal HIV incidence and prevalence, ART coverage and virologic suppression from published data (Table 1). Data informing the natural history of HIV, ART efficacy, EID cascade and HIV‐related costs were also taken from published literature (Table A5) [14, 15]. Selected base‐case model input parameters Abbreviations: Assum., assumption; DTG, dolutegravir; EFV, efavirenz; LPV/r, ritonavir‐boosted lopinavir; NAAT, nucleic acid amplification test; PI, protease inhibitor; SD, standard deviation; SOC, standard of care. We assumed that 86% of HIV‐exposed infants received SOC prophylaxis in all three countries [57], with 75% and 71% efficacy against intrapartum and postnatal transmission, respectively (Table 1, online Appendix; Table A1) [60, 66]. There are no published efficacy studies of bNAb prophylaxis among children. However, in the Antibody Mediated Protection (AMP) studies, HIV incidence with VRC01‐sensitive isolates was 75% lower among adults who received VRC01 than those who received a placebo [6]. Potent bNAbs alone or in combination have neutralizing activity for 84–92% of isolates in multiclade panels [10]. Therefore, we conservatively assumed an average efficacy of 60% for the hypothetical bNAb product against both intrapartum and postnatal transmission, by assuming that bNAb prophylaxis would lead to an approximate 75% reduction in transmission from the estimated 84% of bNAb‐susceptible circulating HIV strains. In the base case, we applied bNAb efficacy equally to high‐risk and non‐high‐risk infants in the HIVE strategies, varying it for each cohort separately in sensitivity analyses. In the base case, we assumed that the bNAb prophylaxis effect duration is 3 months, based on pharmacokinetic studies [7, 8]. Our assumed bNAb cost of $60/dose was driven primarily by estimated production costs (using the highest end of estimates to reflect potential bNAb combination products), but also includes cold‐chain, overhead and personnel costs, based on costs for vaccine delivery (see online Appendix; Table A2) [12]. We conducted univariate and multivariate sensitivity analyses to test the robustness of our findings to changes in key parameters. Given the uncertainty around bNAb uptake, efficacy, effect duration and costs, these inputs were varied widely in univariate sensitivity analyses. We also investigated the influence of uptake and efficacy of WHO‐recommended SOC prophylaxis; recognition of infants’ high‐risk status; breastfeeding duration; maternal retention in care and virologic suppression; vertical transmission risks; EID uptake; paediatric ART efficacy; and ART costs. In multivariate analyses, we simultaneously varied bNAb efficacy and costs to identify which strategies would be cost‐effective across a wide array of hypothetical bNAb characteristics at both the 20% and 50% GDP per capita thresholds.