Probability of vertical HIV transmission: A systematic review and meta-regression

Background Eliminating HIV vertical transmission (VT) is a global priority and monitored by estimating paediatric HIV infections with the Joint United Nations Programme on HIV/AIDS supported Spectrum AIDS Impact Module (Spectrum-AIM). Recent innovations in antiretroviral therapy (ART) service delivery models and first-line regimens aimed to reduce VT probabilities. We conducted a systematic review and meta-analysis to estimate VT probabilities by maternal immunologic and treatment status. Methods We combined an updated systematic review with previous data in meta-regression models to estimate VT probabilities and determinants. We searched multiple databases for peer-reviewed English-language studies from all regions published between January 2018 and February 2024 with VT data stratified by maternal immunologic or treatment status from randomized trials, cohort, or observational studies. Four meta-regression models estimated VT probabilities. We assessed model sensitivity and compared estimates to Spectrum-AIM’s previous results. Finally, we fit a meta-regression model to assess the association of ART class and initiation timing on viral load suppression (VLS) at delivery. Findings The updated review identified 24 new studies, yielding 110 total studies included in meta-regression analysis. For women not receiving ART, higher CD4 was associated with lower odds of perinatal VT (odds ratio [OR] 0.80 (95% CI: 0.75-0.84) per 100 CD4 cells/ µL increase). For pregnant women on ART, each additional week on ART before delivery reduced odds of VT by 5.6% (3.2%–7.0%). The odds ratio of perinatal VT among pregnant women initiating integrase inhibitor-based ART 20 weeks pre-delivery was 0.36 (0.14–0.94) compared to those initiating non-nucleoside reverse transcriptase inhibitors (NNRTI)-based ART. This association was confounded by study region. Odds of VLS were lower when ART was initiated late in pregnancy (OR: 0.37 (0.21-0.68) for the reference regimen (NNRTI)), without significant difference by ART regimen. Interpretation VT probability varies by maternal immunologic stage, treatment regimen, and timing of treatment initiation. These estimates have been incorporated into Spectrum-AIM for UNAIDS 2025 HIV estimates. Earlier ART initiation is associated with higher odds of VLS at delivery. Further evidence is needed on the effects of recent ART innovations on VT outcomes. Funding NIH, UNAIDS, and UKRI RESEARCH IN CONTEXT Evidence before this study Three previous systematic reviews have assessed vertical transmission probability to inform assumptions for modelled estimates of perinatal and breastfeeding vertical HIV transmission with the Spectrum AIDS Impact Module (Spectrum- AIM). The most recent systematic review, conducted in 2018, calculated vertical transmission probabilities as the weighted average of literature published before 2018. Since 2018, dolutegravir has been widely adopted as first-line antiretroviral for all people living with HIV, differentiated service delivery models for pregnant and breastfeeding women have aimed to improve care engagement, and immediate lifelong ART initiation following HIV diagnosis has become standard for all people living with HIV. These biomedical and service delivery interventions aim to increase the proportion of pregnant and breastfeeding women who are virally suppressed before delivery, but evidence is limited about their impact on vertical transmission outcomes. To identify any previous systematic review and meta- regression studies addressing vertical transmission rates and impacts of these innovations therein, we searched PubMed from database inception to 4 January 2025 for published articles using the search terms (“vertical transmission” OR “mother-to-child transmission” OR “MTCT” AND “HIV” AND “systematic review” AND “meta-regression” OR “meta-analysis”), resulting in 27 relevant articles. No articles estimated vertical transmission probability stratified by prevention of vertical transmission intervention. Articles that estimated cumulative vertical transmission probabilities were restricted to a country or a region. Added value of this study This study systematically reviewed literature published before 8 February 2024 on HIV vertical transmission probabilities stratified by maternal clinical or treatment status. It is the first analysis using meta-regression to estimate vertical transmission probabilities that are compatible with Spectrum-AIM. Compared to previous analyses, the meta-regression approach improves accounting for between-study heterogeneity and quantifies statistical uncertainty associated with the vertical transmission probabilities used in Spectrum-AIM. Overall patterns of relative vertical transmission probabilities were similar to previous reviews, but we estimated marginally higher transmission probabilities among breastfeeding women. Initiation of antiretrovirals before the second trimester of pregnancy was associated with higher odds of maternal viral load suppression at delivery. The systematic review identified limited new data about impacts of new first-line drug regimens or service delivery models outside of high-income settings. Implications of all the available evidence When implemented in Spectrum-AIM, the vertical transmission probabilities estimated here result in more infections during breastfeeding than previous estimates. This underscores the importance of testing HIV-exposed infants for final HIV status outcome after breastfeeding has ended. Additionally, our results show the importance of initiating antiretrovirals as early as possible during or before pregnancy to preserve the woman’s health and achieve viral suppression by delivery to reduce the probability of vertical transmission. Further evidence is needed on the effect of integrase inhibitor-based regimen on vertical transmission in diverse global regions.