Azithromycin (AZ), a pregnancy class B macrolide, is a derivative of erythromycin exhibiting bacteriostatic activity by binding to the 50S ribosomal subunit and interfering with protein synthesis. AZ is a broad-spectrum antibiotic affecting aerobic and facultative Gram-negative and Gram-positive bacteria, genital Mycoplasma and Ureaplasma, and some anaerobes ( Peters et al., 1992; Subramaniam et al., 2019). Typically, the antibiotic is used to treat diseases such as respiratory infections, sexually transmitted infections (STIs), and enteric infections caused by a variety of bacterial species ( Heidary et al., 2022).

During pregnancy and parturition, macrolides together with penicillins and cephalosporins are frequently used to prevent or treat infections. AZ, compared to other macrolides (e.g., erythromycin), has additional benefits which may make it advantageous to use more widely for labor and delivery, particularly in low- and middle-income countries (LMIC). These benefits include better pharmacokinetics with a long elimination half-life (68 hours), a high tissue absorption resulting from minimal metabolism by human cytochrome P450 (CYP) enzymes, and additional immunomodulatory and anti-inflammatory properties ( Antonucci et al., 2022; Fohner et al., 2017; Tita et al., 2016; Tita et al., 2009). Since AZ is a broad-spectrum antibiotic, it possesses better coverage against bacteria that are commonly found in polymicrobial infections after labor. Additionally, AZ can rapidly transport across the placenta to provide effective concentrations in the cord blood and amniotic fluid and is found at sustainable levels in breast milk for 48 hours thus providing not only maternal protection but also neonatal protection against infection ( Sutton et al., 2015).

The use of prophylactic antibiotics including macrolides has been shown to reduce infections in preterm premature rupture of membranes (PPROM) and cesarean deliveries ( Keenan et al., 2018; Roca et al., 2016; Tita et al., 2016; Tita et al., 2023). Oral AZ was recognized to have great potential to prevent maternal and neonatal death/infections because of its ease of administration and spectrum activity. NICHD’s Global Network recently completed a trial of a single intrapartum oral dose of AZ versus placebo in eight LMICs (the Azithromycin Prevention in Labor Use study [A-PLUS]) showing a significant decrease in maternal and neonatal infection ( Tita et al., 2023). Other previous randomized AZ trials in LMICs including the Gambia and the Macrolides Oraux pour Réduire les Décès avec un Oeil sur la Résistance (MORDOR) trials have also shown successful reduction in infection of participants who received the drug ( Keenan et al., 2018; Roca et al., 2016). Although much has been done to show the beneficial effects of AZ, more research is needed to address the safety of the drug. One major concern which needs to be considered is antimicrobial resistance, especially resistance to AZ as its long half-life results in extended exposure.

Globally, there is concern regarding the increasing use of AZ. One contributing factor is the availability of this antibiotic and others in community pharmacies without the use of a prescription ( Auta et al., 2019). Additionally, during the COVID-19 pandemic, consumption of AZ increased when it was believed that the antibiotic effectively treated COVID-19 ( Kournoutou & Dinos, 2022). Unfortunately, these actions have resulted in a rise of AZ resistance among various bacterial species, with the greatest concern for species which can only be effectively treated by a macrolide (e.g., Campylobacter spp.). In South Asia, there are reports of AZ-resistant Salmonella typhi and other Salmonella spp. ( Heidary et al., 2022; Hooda et al., 2019). In other parts of the world including Europe, South America, Asia, and Africa, additional species have been reported with increasing rates of AZ-resistance. These include Streptococcus spp., Haemophilus influenzae, Chlamydia trachomatis, Campylobacter spp., Legionella pneumophila, Treponema pallidum, and Neisseria gonorrhoeae ( Derbie et al., 2020; Heidary et al., 2022).

In the MORDOR and the Gambia trials, AZ resistance was also examined. The MORDOR study, which involved a mass drug administration of AZ for the reduction of childhood mortality, showed an increase in community-level AZ resistance in those who received the drug. However, the levels of resistance declined or returned to baseline when the mass distribution stopped ( Doan et al., 2019; Haug et al., 2010; O'Brien et al., 2019). The Gambia study ( Roca et al., 2016) demonstrated that a single oral dose of AZ during labor resulted in an initial higher prevalence of AZ-resistant Staphylococcus aureus among women and their babies four weeks after treatment but the increase subsequently waned by 12 months, and there was no induction of resistance towards other antibiotics by Streptococcus pneumoniae or S. aureus ( Bojang et al., 2018; Roca et al., 2016). Although this study suggests that a single prophylactic dose strategy is less likely to induce long-term antibiotic resistance, further evaluation is required including the effects of AZ on other organisms as well as the microbiome. Furthermore, these studies addressed community-level AZ resistance using surveillance methods but did not address resistant infections.

Therefore, to describe the impact of a single large dose of prophylactic AZ in mothers and neonates on antimicrobial resistance, we planned a prospective sub-study to the A-PLUS trial. This sub-study will monitor antimicrobial resistance of maternal and neonatal infections up to 42 days postpartum and relevant commensal target species, Streptococcus pneumoniae and Staphylococcus aureus from the nasopharynx and Escherichia coli from the rectum, over time up to 12 months postpartum. Collectively, we will be able to assess whether immediate resistance to AZ is acquired by clinical isolates and by the commensal target species following the prophylactic dose and if resistance persists over time.

The A-PLUS AMR sub-study was conducted under the auspices of the A-PLUS trial, conducted by the Global Network for Women’s and Children’s Health Research (Global Network). The trial was overseen by the Global Network’s Steering Committee, comprised of 2 principal investigators (PIs) from each of the 8 participating sites and the PI at the data coordinating center. In addition, there was a A-PLUS Working Group and an AMR Sub-study Working Group, comprised of the lead PIs and DCC investigators, respectively, which developed the analytic plans and provided recommendations to the Steering Committee. Each study site had a lead investigator and study coordinator who oversaw conduct of the research at their site, with support from the DCC and Working Groups. An independent Data Safety and Monitoring Board comprised of technical experts in infectious disease, obstetrics, pediatrics and statistical methods, appointed by the National Institute for Child Health and Human Development (NICHD) reviewed the study enrollment, safety, and trial and sub-study progress. An interim analysis was planned to evaluate the efficacy at 70% of enrollment.

This A-PLUS prospective sub-study is designed to investigate two perspectives of AZ resistance: 1) the resistance phenotype of bacteria causing clinical infections and 2) the surveillance of target commensal bacteria over time and the resistome. In addition to AZ resistance, we are also assessing the microbiome as an indirect potential effect of altered microbial communities not necessarily related to antimicrobial resistance; the methods and results, however, will be described elsewhere. To achieve an understanding of all AZ resistance perspectives, the study will perform clinical monitoring and culturing of bacterial infections and obtain nasopharyngeal (NP) and rectal swab samples from the mother and the neonate over time, with participants from the AZ and placebo-assigned groups. The clinical monitoring and serial surveillance approaches will provide:

The target study population is the same as the primary trial ( Hemingway-Foday et al., 2023; Tita et al., 2023). Pregnant women >28 weeks gestion with a live fetus with a planned vaginal delivery at one of the pre-identified health facilities at eight Global Network sites will be screened by research staff. Women with evidence of chorioamnionitis or other infection requiring antibiotic therapy at the time of enrollment, active use of azithromycin or other macrolide, or any contraindication to macrolides will be excluded. Further details of the eligibility criteria are described elsewhere ( Hemingway-Foday et al., 2023). Health facilities include hospitals and health centers where women routinely deliver. Following randomization, all women received standard clinical care, including antibiotic use, as per the local standards. In addition to routine care, research staff may contact women by phone or home visits to try to ensure high follow-up compliance with study visits.

All microbiology laboratories that support the health facilities including hospitals and health centers at each of the sites ( Table 1) have the basic clinical microbiology capabilities including culture, microorganism identification using biochemical and microscopic methods, and antimicrobial susceptibility testing. This study will build on these basic microbiology capabilities by incorporating microbial screening methods for the selection of specific commensal target species and presumptive AZ resistant isolates. Clinical capabilities will also be expanded for the appropriate collection of samples from infection sites, and the collection of NP and rectal swabs using aseptic techniques. To ensure successful execution of these expanded capabilities, all sites will be provided with detailed standard operating procedures (SOPs) and training sessions which will include on-site hands-on training, virtual training, and training videos. Additionally, quality assurances will be implemented to ensure that all procedures are performed proficiently and consistently at each of the sites providing high quality samples and results. These include the integration of quality controls in the SOPs for both clinical and laboratory tasks and the monitoring of procedures to verify the quality of performance by the sites. Additionally, bacterial isolates from all sites will be validated by the University of Alabama Birmingham to ensure correct identification and susceptibility assessment for AZ. Any inconsistent results or concerns and discrepancies observed will be addressed with additional training.

The clinical monitoring for infection will use the same sample set and monitoring timeline as the primary trial ( Hemingway-Foday et al., 2023). Participants infected by bacterial species will undergo a susceptibility test to determine the minimum inhibitory concentration (MIC) of AZ. The DCC will conduct quality and consistency checks to help ensure the accuracy of the data.

Capacity building for improving sample collection techniques will be done to prevent exogenous contamination during the collection process. Criteria for the collection of clinical samples to prevent exogenous contamination will be followed, including the following:

Site-specific routine clinical microbiological practices will be performed for the isolation and identification of dominant bacterial pathogens found in the various infection sites. These standard practices include culture using various selective and/or differential media (e.g., blood agar, Mannitol salt agar, or MacConkey agar). After 24 – 72 hours of incubation, the dominant organisms (up to three) on each culture plate will be selected for identification according to the routine method at each site, which may include observations of bacterial growth phenotypes (e.g., colony morphology, or variant growth on differential media), microscopic observations using various staining techniques (e.g., Gram stain, India ink, or Acid fast stain), performance of biochemical assay tests (e.g., catalase, indole, or cAMP), and antibody-based test methods (e.g., latex agglutination test).

These dominant isolates will also be assessed for antibiotic susceptibility to AZ and other clinically relevant antibiotics, using standardized procedures and materials such as the disk diffusion method or a minimal inhibitory concentration (MIC) method (e.g., microdilution, agar dilution, Epsilometer test [Etest]). Each bacterial isolate will be interpreted as susceptible, intermediate, or resistant to the challenge antibiotic using the recommended breakpoints set by the Clinical and Laboratory Standard Institute ( CLSI, 2024).

The identified pathogens will be summarized overall by treatment arm as well as separately by region (Asia, Africa, Central America), mother vs. infant, and by infection site. The susceptibility vs. resistance to AZ of relevant pathogens will also be summarized overall by treatment arm as well as by region, pathogen, infection site and by mother vs. infant. These analyses will be descriptive in nature but dependent on having sufficient data to run models, comparisons of treatment arms may be obtained including estimating relative risk estimates between treatment arms for pathogen presence and/or resistance. Full details of any such comparisons will be described in a data processing and statistical analysis plan.

All clinical data will be entered into an electronic data capture system (REDCap 14.3.14; https://projectredcap.org/), developed by the DCC and hosted at each study site. Data consistency, quality and range checks will be performed at each site and centrally. Details of the data management plans are included in the methods paper ( Hemingway-Foday et al., 2023).

A random subset of maternal-infant dyads at each site as selected by the DCC, enrolled prior to randomization, will have NP and rectal specimens obtained and monitored serially (at baseline [0–1 day], 1 week, 6 weeks, 3 months, 6 months, and 12 months) to assess the development of AZ resistance of target bacterial flora found in the NP and the rectum. The time points listed correspond to days postpartum which will also correlate to days when the prophylactic dose of AZ (two 500 mg film-coated tablets; Kern Pharma; Barcelona, Spain) is administered since administration will happen at the time of delivery in the site-specific health facilities. Baseline sampling will be taken at a time before or a day after the AZ is administered to the mother. Following the AZ administration, an early time point (i.e., 1 week), mid time points (i.e., 6 weeks and 3 months), and late time points (i.e., 6 months and 12 months) will be assessed to determine if AZ resistance is potentially acquired and how long the resistance persists.

Bacterial collection will be performed following an SOP using sterile flocculated swabs. A mini-tip nasopharyngeal swab will be used to collect the nasopharyngeal flora in addition to the anterior nares and a standard swab will be used for the rectum. The swabs will be placed in a tube with 1 mL of Amies or Stuarts transport medium. The sampling scheme for this monitoring study is shown in Table 2. A similar scheme will be used for collection and storage of specimens for microbiome (MB) analysis (see below).

The target commensal bacteria selected include Streptococcus pneumoniae and Staphylococcus aureus from the nasopharynx and anterior nares, respectively, and Escherichia coli from the rectum. Both S. pneumoniae and S. aureus are clinically relevant species in which AZ is used to treat infections caused by these bacteria. Additionally, there is a worldwide increase in AZ resistance of these species, and it is important for us to monitor these bacteria to determine whether a prophylactic dose of AZ will contribute to their resistance. E. coli is a representative species abundantly found in the rectum and will be monitored for Gram-negative resistance even though AZ is not a primary drug for most E. coli infections.

Assuming that the primary endpoint of interest is AZ resistance to at least one organism from any flora at each time point, samples from a total of 848 maternal-infant dyads will need to be monitored to estimate an overall prevalence of 10% +/- 2% with 95% confidence. Furthermore, this sample size will allow over 80% power to detect at least a 2-fold increase in AZ microbial resistance with AZ (RR=2) compared to 6–7% in placebo group. We will also be able to compare the number of organisms identified with resistance to AZ between randomized study groups. To account for an assumed rate of 15% for individuals declining consent to store samples for future testing, the final sample size goal will be 1000 maternal-infant dyads total. This corresponds to approximately 125 dyads per GN country for the duration of the study. Enrollment may vary across countries, but no individual country will enroll more than 25% of the planned sample size.

All sites, following the same SOP for the processing of the NP and rectal swabs, will initially plate an aliquot of each swab sample onto a selective and differential agar plate. The NP swabs will be plated onto gentamicin blood agar (5 µg/mL of gentamicin; 5% blood) for the isolation of S. pneumoniae and onto Mannitol salt agar (Fisher Scientific; Waltham, MA) for the isolation of S. aureus. The rectal swabs will be plated onto Brilliant E. coli and coliforms agar (Fisher Scientific; Waltham, MA) for the isolation of E. coli. From each agar plate, up to 50 presumptive isolates with the correct colony phenotypes will be picked and patched onto Mueller Hinton or blood agar plates with high concentrations of AZ to screen for resistant isolates. The concentration of AZ is determined based on the recommended MIC breakpoints for resistance found in the CLSI. Since the CLSI does not contain MIC breakpoints for E. coli, we will use the breakpoints specified for Enterobacteriaceae. Identification of the presumptive resistant isolates that grow will be confirmed using standardized biochemical assays and the AZ resistance will be validated using the Etest strips (Liofilchem™ MTS™ Azithromycin; Fisher Scientific; Waltham, MA).

The same subset of dyads for AZ susceptibility monitoring described above will also provide an NP and rectal swab sample for microbiome (i.e., 16S V4 rRNA sequencing and metagenomics) and resistome analyses. This corresponds to a total of 1000 maternal-infant dyads who are randomly selected, consent to participate in the sub-study, and agree to have samples stored for future analysis.

Sample collection will follow the same SOP that will be performed for the AZ susceptibility monitoring, and the swabs will be placed in 1 mL of Zymo DNA/RNA Shield buffer (Zymo Research; Irvine, CA). The samples will be stored at -80°C until ready for DNA extraction, sequencing, and analysis.

The processing and sequencing methods will be described elsewhere.

The AMR sub-study protocol has been reviewed and approved on July 1, 2020 (approval number: IRD-300002323) by the University of Alabama Institutional Review Board and the University of Alabama Ethics Committee at the University of Alabama at Birmingham, which serves as the clinical lead site, and by other relevant ethics committees and regulatory authorities at each individual research site (Kinshasa School of Public Health, DRC; University Teaching Hospital, University of Zambia, Zambia; Moi University, Eldoret, Kenya; INCAP, Guatemala City, Guatemala; Lata Medical Research Foundation, Nagpur, India; KLE Academy for Research and Higher Education’s JN Medical College, Belagavi, India; The Aga Khan University, Karachi, Pakistan; icddr,b, Dhaka, Bangladesh) including the IRB at RTI International, which serves as the Data Coordinating Center. The DMC has reviewed the study protocol and will continue to review throughout the enrollment period. The study was registered (clinicaltrials.gov NCT03871491).

All research staff responsible for obtaining informed consent were trained and certified in the protection of human subjects and the study-specific consent procedures. A model written informed consent form, developed according to the requirements of the OHRP, may be modified by each site to conform to local standards, but the OHRP required elements must be maintained. The research sites will also be responsible for translating the consent form into the appropriate language(s) for their local context. All women will provide informed written consent prior to participation.