Tuberculosis Preventive Treatment for Pregnant People With HIV in South Africa: A Modeling Analysis of Clinical Benefits and Risks

Abstract

Background: Although prior studies of tuberculosis-preventive treatment (TPT) for pregnant people with human immunodeficiency virus (PPWH) report conflicting adverse pregnancy outcome (APO) risks, international guidelines recommend TPT for PPWH.

Methods: We used a microsimulation model to evaluate 5 TPT strategies among PPWH receiving antiretroviral therapy in South Africa: No TPT; 6 months of isoniazid (6H) or 3 months of isoniazid-rifapentine (3HP) during pregnancy (Immediate 6H or Immediate 3HP) or post partum (Deferred 6H or Deferred 3HP). The primary outcomes were maternal, fetal/infant, and combined deaths from causes potentially influenced by TPT (maternal tuberculosis, maternal hepatotoxicity, stillbirth, low birth weight [LBW], and infant tuberculosis). Tuberculosis during pregnancy confers 250% and 81% higher modeled risks of stillbirth and LBW, respectively. In lower-risk or higher-risk scenarios, immediate TPT confers 38% lower or 92% higher risks of stillbirth and 16% lower or 35% higher risks of LBW.

Results: Immediate TPT would minimize deaths among PPWH. When TPT confers higher stillbirth and LBW risks, immediate TPT would produce the most combined maternal and fetal/infant deaths, even with low maternal CD4 cell count and high tuberculosis incidence. If immediate TPT yields a <4% or <20% increase in stillbirth or LBW, immediate TPT would produce fewer combined deaths than deferred TPT (sensitivity analysis range, <2%-22% and <11%-120%, respectively).

Conclusions: If APO risks are below identifiable thresholds, TPT during pregnancy could decrease combined maternal and fetal/infant deaths. Given uncertainty around isoniazid's risks, and the low threshold at which APO risks could outweigh benefits from tuberculosis deaths averted, studies of newer TPT regimens among PPWH are warranted to inform guidelines.

Keywords: HIV; adverse pregnancy outcome; pregnancy; tuberculosis; tuberculosis preventive treatment.

Graphical Abstract

This graphical abstract is also available at Tidbit: https://tidbitapp.io/institutional-portal/clinical-infectious-diseases/tidbits/tuberculosis-preventive-treatment-for-pregnant-people-with-hiv-in-south-africa-a-modeling-analysis-of-clinical-benefits-and-risks/update

Figure 1.

Model-projected deaths among 213 928 pregnant people with human immunodeficiency virus (HIV) and their children in South Africa over 3.5 years, by cause, tuberculosis preventive treatment (TPT) strategy, and assumption about adverse pregnancy outcome (APO) risk. Results are scaled to the annual number of pregnant people with HIV on antiretroviral therapy before pregnancy in South Africa (n = 213 928) from the South Africa Antenatal Survey 2019 [25, 26]. This figure shows causes of death for pregnant and postpartum people with HIV and their children by TPT strategy: No TPT (gray), immediate treatment with 6 months of daily isoniazid (6H) (dark orange), immediate treatment with 3 months of weekly isoniazid-rifapentine (3HP) (dark blue), Deferred 6H (light orange), and Deferred 3HP (light blue). For stillbirth and low birth weight (LBW), results from 3 scenarios are shown: a higher-risk scenario in which Immediate 6H yields more APOs than Deferred 6H and No TPT, as reported in the TB APPRISE randomized controlled trial [8]; a lower-risk scenario in which Immediate 6H yields fewer APOs than Deferred 6H and No TPT, as reported by Kalk et al [11]; and an equivalent-risk scenario in which Immediate 6H yields no difference in APOs compared with Deferred 6H and No TPT. Maternal tuberculosis deaths during pregnancy, fetal demise from maternal death, stillbirths, and LBW deaths are the same in deferred and no-TPT strategies. Maternal deaths from TPT-associated hepatotoxicity are: 4 for Immediate 6H, 2 for Immediate 3HP, 40 for Deferred 6H, and 13 for Deferred 3HP. Fetal demise is due to maternal tuberculosis or TPT-associated hepatotoxicity death during pregnancy. Infant deaths from tuberculosis are: 5 for No TPT, 4 for Immediate 6H, 3 for Immediate 3HP, 4 for Deferred 6H, and 3 for Deferred 3HP.

Figure 2.

Model-projected changes in maternal and fetal/infant deaths over 3.5 years with tuberculosis preventive treatment (TPT) during pregnancy or post partum, compared with No TPT. Results are scaled to the number of pregnant people with human immunodeficiency virus (PWH) on antiretroviral therapy before pregnancy in South Africa [n = 213 928] from the South Africa Antenatal Survey 2019 [25, 26]. This figure shows the number of deaths in TPT strategies compared with No TPT, including pregnant/postpartum PWH and fetal/infant deaths. Fetal/infant results include 3 scenarios: assuming either higher, lower, or equivalent risk of adverse pregnancy outcomes from TPT during pregnancy compared with deferred or No TPT. The only difference in infant deaths between deferred TPT strategies and No TPT is maternal tuberculosis transmissions to infants post partum, which results in 1 additional infant tuberculosis death averted with deferred TPT strategies. Abbreviations: 3HP, 3 months of weekly isoniazid-rifapentine; 6H, 6 months of daily isoniazid.

Figure 3.

One- way sensitivity analysis model results for maternal and fetal/infant deaths over 3.5 years, comparing immediate versus deferred treatment with 6 months of daily isoniazid (6H) under different assumptions about adverse pregnancy outcome (APO) risk associated with tuberculosispreventive treatment (TPT) during pregnancy. Results are scaled to the annual number of pregnant people with human immunodeficiency virus (HIV) on antiretroviral therapy before pregnancy in South Africa (n = 213 928) from the South Africa Antenatal Survey 2019 [25, 26]. Differences in deaths reflect projected deaths with Immediate 6H versus Deferred 6H; negative numbers indicate that Immediate 6H produces fewer deaths than Deferred 6H, while positive numbers indicate that Immediate 6H produces more deaths. A, Maternal deaths. B–D, For fetal/infant deaths, results from 3 scenarios are shown: a higher-risk scenario in which Immediate 6H yields more APOs than Deferred 6H, as reported in the TB APPRISE randomized controlled trial (B) [8]; a lower-risk scenario in which Immediate 6H yields fewer APOs than Deferred 6H, as reported by Kalk et al [11] (C); and an equivalent-risk scenario in which Immediate 6H yields no difference in APOs compared with Deferred 6H (D). Vertical lines represent the base case values, and the width of the bars reflects the lowest and highest projected values within each sensitivity analysis range. The range of values considered in sensitivity analysis are ordered: the value yielding the most favorable result for Immediate 6H is listed first, and the value yielding the least favorable result for Immediate 6H is listed second. Tuberculosis progression indicates progression from tuberculosis infection to tuberculosis disease among pregnant and postpartum people with HIV; 0.25×, 0.5×, and 2× indicate quartering, halving, and doubling the base case value, respectively; and tuberculosis mortality indicates deaths from untreated tuberculosis among pregnant and postpartum people with HIV.

Figure 4.

Sensitivity analysis model results: the maximum relative increase in risk of stillbirth or low birth weight (LBW) under which immediate treatment with 6 months of daily isoniazid (6H) would produce fewer combined maternal and fetal/infant deaths than Deferred 6H. Abbreviation: TPT, tuberculosis-preventive treatment. This figure shows the maximum relative increase in risk of stillbirth (A) or LBW (B) under which Immediate 6H (given during pregnancy) would produce fewer combined maternal and fetal/infant deaths than Deferred 6H (given post partum). Vertical lines represent base case values, and the width of the bars reflects the lowest and highest projected values within each sensitivity analysis range. The range of values considered in sensitivity analysis are ordered; the value yielding the lowest maximum threshold for the relative increase in stillbirth or LBW in sensitivity analysis is listed first, and the value yielding the highest maximum threshold for the relative increase in stillbirth or LBW in sensitivity analysis is listed second. Full results are provided in Supplementary Table 9, and sensitivity analysis results for treatment with 3 months of isoniazid-rifapentine are provided in Supplementary Table 9 and Supplementary Figure 12. Tuberculosis progression indicates progression from tuberculosis infection to tuberculosis disease among pregnant and postpartum people with human immunodeficiency virus (HIV); 0.25×, 0.5×, and 2× indicate quartering, halving, and doubling the base case value, respectively; and tuberculosis mortality indicates deaths from untreated tuberculosis among pregnant and postpartum people with HIV.