Abstract
Background
Despite various studies reporting conflicting risks regarding adverse pregnancy outcomes (APOs) linked to tuberculosis-preventive treatment (TPT) for pregnant individuals living with human immunodeficiency virus (HIV), global health authorities have maintained their recommendation for TPT among this vulnerable population. Notably, the World Health Organization supports TPT, highlighting its importance in pivotal moments of care.
Methods
Utilizing a comprehensive microsimulation model, we thoroughly evaluated five distinct TPT strategies deployed among pregnant people living with HIV receiving antiretroviral therapy in South Africa. These included: No TPT; six months of isoniazid (6H) or three months of isoniazid-rifapentine (3HP) treatment initiated during pregnancy, termed Immediate 6H or Immediate 3HP; and treatments deferred until after childbirth, referred to as Deferred 6H or Deferred 3HP. Critical outcomes assessed included maternal, fetal/infant, and combined mortality stemming from causes that TPT may potentially influence, such as maternal tuberculosis, complications from hepatotoxicity, stillbirths, low birth weight (LBW), and infant tuberculosis infections. The study underscored the significantly heightened risks of stillbirth (250%) and low birth weight (81%) associated with tuberculosis during pregnancy. Furthermore, in various risk scenarios, immediate TPT was found to correlate with a 38% reduction or a 92% increase in stillbirth risks, and a 16% reduction or a 35% increase in low birth weight risks.
Results
Our findings suggest that immediate TPT would substantially reduce mortality among pregnant individuals living with HIV. However, in situations where TPT leads to elevated stillbirth and low birth weight risks, the immediate TPT approach was projected to yield the highest total combined maternal and fetal/infant deaths, particularly notable even at diminished maternal CD4 cell counts coupled with elevated tuberculosis incidence.
Conclusions
If the risks associated with adverse pregnancy outcomes are identifiable and demonstrably low, implementing TPT during pregnancy could effectively decrease overall maternal and fetal/infant mortality rates. Thus, accounting for the uncertainty surrounding isoniazid’s associated risks, coupled with the minimal thresholds at which adverse pregnancy outcome risks may overshadow the benefits conferred by lives saved from tuberculosis, indicates that further exploration of novel TPT regimens among pregnant individuals living with HIV is essential to refine and update clinical guidelines.
Globally, tuberculosis ranks as the primary cause of mortality among individuals living with HIV. Throughout pregnancy and the postpartum period, incidence rates of tuberculosis tend to be higher than during other life stages. Consequently, pregnancy represents a critical opportunity to provide TPT, especially considering elevated mortality risks from tuberculosis and improved access to healthcare observed during this timeframe. However, existing data raise concerns about potential adverse pregnancy outcomes (APOs) that may be linked to TPT, notably stillbirth and low birth weight.
The TB APPRISE randomized controlled trial (RCT) reported an increase in composite APOs among pregnant individuals living with HIV who began six months of isoniazid preventive therapy during pregnancy, compared to those who initiated therapy postpartum. These conclusions contrast with findings from various nonrandomized studies that indicated lower APO risks among those who received isoniazid preventive therapy during pregnancy versus non-receiving counterparts. In light of these findings, the World Health Organization recommends isoniazid preventive therapy for pregnant individuals living with HIV to prevent serious tuberculosis infection since deferment could result in significant vulnerability during a critical healthcare stage.
Our primary aim was to analyze the potential trade-offs inherent in various TPT strategies for pregnant individuals living with HIV in South Africa, a country grappling with a significant burden of HIV and tuberculosis, by considering diverse assumptions surrounding the benefits and risks associated with TPT utilization.
METHODS
Analytic Overview
Using the Cost-Effectiveness of Preventing AIDS Complications (CEPAC) model, which is a recognized microsimulation reflecting the natural history of HIV and tuberculosis alongside their treatment, we estimated the health outcomes attributed to TPT strategies for pregnant individuals living with HIV and their offspring. The evaluation encompassed five treatment strategies: No TPT; a six-month treatment period of daily isoniazid (6H) starting in the fourth month of pregnancy (Immediate 6H); a three-month treatment period employing weekly isoniazid-rifapentine (3HP) commencing in the fourth month of pregnancy (Immediate 3HP); a six-month treatment initiation three months post childbirth (Deferred 6H); and a three-month treatment after childbirth (Deferred 3HP). Given that the majority of pregnant individuals living with HIV in South Africa typically engage with antenatal care starting in the second trimester, our simulations began from the fourth month of pregnancy. The projected outcomes were analyzed over the duration of six months during pregnancy and three years postpartum, yielding a total observation period of 3.5 years. This analysis further scaled results to align with the estimated annual population of pregnant individuals living with HIV in South Africa who commence ART prior to pregnancy, approximating 213,928 individuals. A comprehensive sensitivity analysis was meticulously conducted concerning pivotal model parameters and assumptions.
Given the existing contradictory reports regarding the APO risks associated with TPT, we examined three critical scenarios: one that indicated a higher risk of APOs from TPT delivered during pregnancy versus those administered post-partum (supported by the TB APPRISE RCT and various studies); the second suggesting a lower risk associated with TPT during pregnancy versus no treatment or deferred treatment; and the third proposing an equivalent risk estimation with TPT during pregnancy versus non-administration. Specific adverse pregnancy outcomes identified for this examination included instances of stillbirth and those associated with low birth weight deaths. Special attention was also allocated to account for heightened risks linked with any tuberculosis incident experienced during pregnancy.
Our study concentrated on deaths resulting from various causes likely influenced by TPT interventions, notably maternal tuberculosis, hepatotoxic outcomes in mothers, stillbirth, incidents of LBW, and pediatric tuberculosis. The analysis presents results related only to pregnant individuals living with HIV, those pertaining distinctly to fetal and infant outcomes, along with information combining both maternal and fetal/infant outcomes to facilitate comprehensive understanding (summing maternal and fetal/infant deaths).
Model Structure
The CEPAC model operates on a monthly cycle simulating HIV and tuberculosis disease progression. At the model’s initiation, an initial CD4 cell count is assigned to each simulated individual. This count critically dictates the monthly probabilities for clinical events, such as developing tuberculosis or other opportunistic infections. Notably, individuals may disengage from HIV care, which subsequently influences their viral load and CD4 cell count metrics. For a comprehensive understanding, detailed model guidelines, including flowcharts and state-specific descriptions, are accessible through online sources.
Model Input Parameters for PPWH
We compiled relevant model input parameters illustrated in Table 1, further described here to encapsulate critical aspects.
Table 1.
Parameters for Analysis of Tuberculosis Preventive Treatment Strategies in Pregnant Individuals with HIV in South Africa
| Parameter
. |
Base Case
. |
Rangea
. |
Source
. |
|---|---|---|---|
Table 1.
Parameters for Analysis of Tuberculosis Preventive Treatment Strategies in Pregnant Individuals with HIV in South Africa
| Parameter
. |
Base Case
. |
Rangea
. |
Source
. |
|---|---|---|---|
| Parameter
. |
Base Case
. |
Rangea
. |
Source
. |
|---|---|---|---|
HIV Treatment
Within the model, pregnant individuals living with HIV commence with dolutegravir-based antiretroviral therapy, generally presenting with a mean CD4 cell count of 696/µL, with approximately 76% demonstrating virologic suppression. We assume that all individuals remain engaged in HIV care throughout pregnancy; however, some may disengage postpartum, resulting in discontinuation of TPT or non-initiation of treatment.
Tuberculosis
According to simulation conditions, we commenced with no pre-existing tuberculosis disease, in agreement with South African guidelines counseling screening for tuberculosis at initial antenatal care visits for individuals living with HIV. The baseline scenario indicates that 35% of participants are at risk due to latent Mycobacterium tuberculosis infection, anticipating a progression risk that depends on individual CD4 counts. Furthermore, we modeled a heightened incidence of tuberculosis disease during pregnancy and the first six months postpartum, where risk ratios could be approximately 1.3 and 1.9 times that of nonpregnant adults.
TPT Regimens
In the base scenario, approximately 79% of pregnant individuals living with HIV engaging in care initiate TPT. Each month, about 1.84% of those starting TPT subsequently discontinue. The efficacy of 6H was ascertained from the TEMPRANO RCT amid nonpregnant individuals living with HIV, yielding a post-adjustment efficacy of 43%, corresponding with recognized estimates. For the 3HP regimen, the derived efficacy lay at approximately 49%. It is essential to note that while we accepted similar efficacy levels during pregnancy and postpartum for treatment protocols, actual TPT effectiveness and completion rates often decline postpartum due to heightened disengagement from HIV care and increased discontinuation of TPT.
Fetal/Infant Health Outcomes
The fetal and infant health outcomes were tailored based on the cohort size at the model’s initiation. We incorporated considerations of HIV vertical transmission, which subsequently escalates the risk of tuberculosis in infants.
APO Risks
Focusing intently on two principal types of adverse pregnancy outcomes—stillbirth (or spontaneous abortion) and low birth weight—our analysis established a contextual foundation for further assessments. For instances of low birth weight, we set an applied death risk probability at 8.1% within the first month.
We determined risk ratios for stillbirth and low birth weight by comparing TPT-conferred outcomes during pregnancy against a baseline of no TPT implementation across three diverse scenarios: the higher-risk scenario indicated by the TB APPRISE RCT, the lower-risk scenario per findings of Kalk et al, and an equivalent-risk scenario described by earlier studies. Notably, due to limitations within available data, we assumed no distinguishable differences in AP risks between the 6H and 3HP treatment regimens.
Tuberculosis
It is critical to consider that any incidence of tuberculosis in infants is presumed to arise from maternal transmission occurring postpartum rather than in utero or from other sources. Prior studies suggested a 6.8%–9.2% probability of developing tuberculosis disease among young children due to proximate contact with individuals exhibiting tuberculosis, adjusted for the infant’s HIV status. Mortality risk among young children contracting tuberculosis is noted at 1.36%.
Sensitivity and Uncertainty Analysis
Sensitivity and Scenario Analysis
Our analyses varied a myriad of parameters pertinent to both pregnant individuals living with HIV and their infants to establish sensitive analytical results. The parameters, alongside their ranges, are elucidated in Table 1.
Threshold Analysis
We identified the upper limits of TPT-associated risks relative to adverse pregnancy outcomes necessary for immediate TPT to generate fewer combined deaths than Deferred 6H. This threshold analysis was conducted by varying stillbirth and LBW risks independently—where changes in the risk ratio for stillbirth maintained the LBW ratio at 1 and vice versa—followed by simultaneous assessments. This process allowed us to perform a two-way sensitivity analysis that concurrently modified mean initial CD4 cell count and TPT efficacy.
Ethics Statement
Ethical approval for this comprehensive study was secured from the Mass General Brigham Human Research Committee under protocol 2014P002708. Because all data employed in the model were derived from published or publicly available sources, no individual participants were involved in this study.
RESULTS
Base Case
Maternal Outcomes
Within our model projections involving 213,928 pregnant individuals living with HIV in South Africa, the anticipated deaths due to tuberculosis and TPT-related hepatotoxicity over a span of 3.5 years were lowest with Immediate 3HP (811 deaths, or 0.38%), followed by Immediate 6H (885 deaths, or 0.41%), Deferred 3HP (970 deaths, or 0.45%), Deferred 6H (1036 deaths, or 0.48%), and the largest mortality count occurred in the No TPT category (1207 deaths, or 0.56%). These results are visually summarized in accompanying figures.
Model-projected deaths for the 213,928 pregnant individuals living with HIV and their children in South Africa over the 3.5 years are categorized by cause, delineated across various TPT strategies and assumptions regarding adverse pregnancy outcome risks. The presented results are correspondingly scaled to account for the annual demographic of pregnant individuals living with HIV receiving antiretroviral therapy before pregnancy in South Africa.
For stillbirth and low birth weight (LBW), results from three diverse scenarios are illustrated: a higher-risk scenario indicating that Immediate 6H incurs more APOs than Deferred 6H and No TPT, as evidenced in the TB APPRISE randomized controlled trial; a lower-risk scenario where Immediate 6H results in fewer APOs than Deferred 6H and No TPT; and an equivalent-risk scenario showing no observable differences in APOs between Immediate 6H and its deferred counterparts. Maternal deaths derived from TPT-associated hepatotoxicity are quantified at 4 for Immediate 6H, 2 for Immediate 3HP, 40 for Deferred 6H, and 13 for Deferred 3HP. Fetal demise attributed to either maternal tuberculosis or mortality resulting from TPT-associated hepatotoxicity during pregnancy was also considered. Additionally, infant deaths occurring due to tuberculosis were counted: 5 for No TPT, 4 for Immediate 6H, 3 for Immediate 3HP, 4 for Deferred 6H, and 3 for Deferred 3HP.
Fetal/Infant Outcomes
Our comprehensive scenarios included considerations for background nontuberculosis/non-TPT associated APOs. Within these frameworks, there are projected occurrences of 7709 stillbirths (3.60%) and 1519 LBW-associated deaths (0.71%) among individuals assigned to deferred TPT and No TPT categories. In the context of the modeled higher-risk scenario, where TPT treatments may amplify APO risks, the Immediate 6H and Immediate 3HP protocols yield 12,943 and 12,941 stillbirths (6.05%) coupled with 1909 LBW-associated fatalities (across both treatment regimens). In contrast, under the lower-risk scenario parameters, results project 5439 stillbirths alongside 5435 LBW-associated deaths (0.62%) for both regimens. Moreover, in the equivalent-risk scenario, 7687 and 7684 stillbirths (3.59%) emerge, along with 1518 LBW-associated fatalities, slightly lowering incidence rates compared to deferred TPT and No TPT, primarily from reduced maternal tuberculosis incidence. Notably, although maternal tuberculosis transmissions to infants post-partum are statistically infrequent with slightly more favorable outputs observed across all TPT strategies as compared with No TPT.
Combined Maternal and Fetal/Infant Outcomes
A cross-comparison across modeled TPT strategies indicates that each approach results in fewer maternal fatalities within 3.5 years than the No TPT strategy. When reviewing maternal deaths together with fetal and infant deaths, we observe an incremental increase, whereby 5251 (1.23%) and 5163 (1.21%) more combined deaths arise from Immediate 6H and Immediate 3HP respectively in the higher-risk scenario relative to No TPT. Conversely, the lower-risk framework reflects reductions of about 2827 (0.66%) and 2917 (0.68%) in combined fatalities for Immediate 3HP and Immediate 6H respectively when compared with No TPT. Similarly, within the equivalent-risk scenario, reductions of 395 (0.09%) and 485 (0.11%) were observed with Immediate 3HP and Immediate 6H when compared to No TPT respectively, reinforcing the conclusion that Deferred 6H and Deferred 3HP yielded respective declines of 172 (0.04%) and 238 (0.06%) in combined mortality when set against No TPT.
Model projections showcase alterations in maternal and fetal/infant fatalities over a span of 3.5 years related to tuberculosis preventive treatment administered during pregnancy or postpartum, thereby juxtaposing various scenarios against the No TPT strategy. The cumulative results reflect the total death counts from intended TPT strategies compared against absence of treatment, incorporating deaths across both pregnant individuals living with HIV and infant-related fatalities. The isolated results for infants span across three outlined scenarios to evaluate potential risks associated with adverse pregnancy outcomes stemming from TPT, juxtaposed against deferred and No TPT contexts. The marginal differential in infant deaths between deferred TPT strategies and the No TPT condition primarily stems from maternal tuberculosis transmissions post-partum.
Sensitivity and Uncertainty Analysis
Sensitivity and Scenario Analysis
Through judicious one-way sensitivity analyses, we compared projected maternal and fetal/infant fatalities over a time frame of 3.5 years between immediate and deferred treatment utilizing 6 months of daily isoniazid (6H), while considering a range of assumptions pertaining to the risks associated with adverse pregnancy outcomes arising from tuberculosis-preventive treatment during pregnancy. These analyses were correspondingly scaled to reflect the population metrics for pregnant individuals living with HIV receiving antiretroviral therapy prior to pregnancy, specifically targeting the South African demographic.
Threshold Analysis
In context, should Immediate 6H elevate the relative stillbirth risk, it would thereby surpass the threshold to yield fewer combined maternal and fetal/infant fatalities than Deferred 6H over the designated 3.5 years. The analogous thresholds for Immediate 3HP would present in equal measure, with the TB APPRISE RCT evidencing increased risks of 92% for stillbirth and 35% for low birth weight. Notably, opportunities for effective treatment markedly amplify with declining initial maternal CD4 cell counts, showcasing how elevated associated risks from TPT may still yield lower combined mortalities for immediate TPT as compared to deferred strategies. Despite the aforementioned, even with a critically low CD4 cell count of 50/µL, recorded relative increases in the stillbirth risk concerning TPT stand at a maximum of 22% for Immediate 6H, remaining beneath the risks reported from the TB APPRISE findings. Importantly, minimizing both the maximum relative increases in stillbirth and low birth weight risks provides further nuanced insights into the dynamics of TPT applied during acute care contexts.
DISCUSSION
The implications from our findings are pivotal for shaping clinical guidelines and directing future research endeavors linked to this field. WHO guidelines advocate for TPT in pregnant individuals living with HIV, despite conflicting trial outcomes demonstrating heightened risks for adverse pregnancy outcomes. Presently, South African medical guidelines endorse a comprehensive regimen for 12 months of isoniazid among this demographics although 3HP may emerge as the preferred treatment option moving forward. Analysis has concluded that immediate TPT may contribute to higher combined deaths than deferred options given observations from TB APPRISE; however, it is noteworthy that a substantial number of TB APPRISE participants were engaged in efavirenz-based ART compared to the now-advocated dolutegravir-based ARV regimen. The interaction between isoniazid and efavirenz may yield results divergent from those experienced with the dolutegravir-based regimen. This uncertainty illuminates the necessity of incorporating pregnant individuals within tuberculosis prevention trials alongside routine pharmacovigilance scrutiny to safeguard maternal and fetal health outcomes adequately.
While the recommendation positions TPT provision during pregnancy solely among individuals with latent tuberculosis infection (TBI), challenges persist around the effective diagnosis of TBI within the context of individuals living with HIV, compounded by evidence suggesting no efficacy variance in TPT outcomes contingent on tuberculin skin test results. However, our conclusions remained consistent throughout when restricting TPT administration to those identified with TBI. Compelling points supporting TPT implementation during pregnancy include significantly elevated tuberculosis risks and associated mortality outcomes, with promising engagement rates observed during COBP visits, potentially optimizing TPT uptake and adherence alongside improved healthcare intersections.
Adverse engagements yield concerning trajectories, as only a modest proportion of individuals living with HIV globally access TPT interventions despite recommendations. The advancement of an effective TPT regimen that can be recommended and possibly completed during pregnancy—at a time marked by frequent healthcare visits—may prove integral in increasing treatment uptake. The 3HP regimen, receiving endorsement from WHO, may effectively satisfy this need. Preliminary observations regarding safety and dosing in pregnant individuals yield encouraging safety profiles, with the validation of forthcoming controlled trials poised to enhance our understanding of TPT’s applicability in these contexts.
Thus, our findings reflect that while TPT in pregnancy could substantially mitigate mortality rates among individuals living with HIV, the globally advocated strategy may inadvertently escalate mortality figures if risks associated with adverse pregnancy outcomes align closely with those disclosed in the TB APPRISE dataset. However, contemporary iterations of TPT regimens tailored for pregnant individuals undergoing integrase inhibitor-based ART may reframe this dynamic, necessitating robust studies targeting TPT usage amidst modern ART to refine guidelines and support informed decision-making for care among pregnant individuals.
