Impact of Intrapartum Azithromycin on the Carriage and Antibiotic Resistance of Escherichia coli and Klebsiella pneumoniae in Mothers and their Newborns: a sub-study of a Randomized Double-Blind Trial Conducted in The Gambia and Burkina Faso

Background: Limited data exists on effects of intrapartum azithromycin on prevalence of carriage and antibiotic resistance of Enterobacterales.


INTRODUCTION
Efforts to reduce global neonatal mortality rates have led to a 50% decrease, from 36.6 to 17.5 per 1000 live births between 1990 and 2019 [1].Nevertheless, progress varies across regions [2].Over the same period, neonatal mortality rates in sub-Saharan Africa decreased by 26%, currently representing 43% of global neonatal deaths [1].Neonatal sepsis, a major contributor to neonatal mortality [3], is often caused by Staphylococcus aureus, Escherichia coli and Klebsiella spp., with varying prevalence in different African sub-regions [4,5].For example, S. aureus sepsis is more prevalent in West Africa, while Klebsiella spp.sepsis is more common in Central and South Africa [4].Maternal vaginal S. aureus colonization, which is correlated with neonatal colonization and subsequent disease, is estimated to be 16% in West Africa 29%, in Central Africa and 2-8% in East Africa [6][7][8][9].
Azithromycin, a second-generation macrolide antibiotic, exhibits broad-spectrum activity against gram-positive and some gram-negative bacteria [10].Its oral administration results in rapid absorption reaching peak concentrations in blood or tissues within 2-3 hours [11], becoming a potential prophylactic antibiotic for preventing neonatal and maternal infections.
Recent double-blinded randomized trials, PregnAnZI-2 and A-PLUS, explored the use of azithromycin to decrease neonatal sepsis and mortality across nine African and Asian countries [12,13].Although no reduction in neonatal sepsis and mortality was observed, a significant impact in reducing maternal infections [12], including puerperal sepsis [13], was noted.The PregnAnZI-2 trial, conducted in West Africa, also reported a reduction in neonatal infections and a lower rate of prescribed antibiotics during the neonatal period [12].
Our earlier research showed that this intervention reduces gram-positive bacterial colonization in mothers and newborns throughout the neonatal period, including S. aureus [6].Despite this, a temporary increase in azithromycin-resistant S. aureus lasting between 1-12 months was observed [14].Additionally, intrapartum azithromycin lowered carriage of Streptococcus pneumoniae and Groups A and B Streptococcus, without increasing antibiotic resistance [6,15].Data on the effect of intrapartum azithromycin on carriage and antibiotic resistance of gramnegative bacteria causing neonatal sepsis, E. coli and K. pneumoniae, are scarce [16].It is important to evaluate the impact of the intervention on these two gram-negative bacteria due to their role in neonatal sepsis and their rising rates of multidrug resistance which severely limits treatment options [17].The study presented here aims to determine the effect of intrapartum azithromycin on the prevalence of carriage and antibiotic resistance of E. coli and K. pneumoniae among mother-infant pairs from The Gambia and Burkina Faso.

Trial overall design
PregnAnZI-2 was a phase III double-blind placebo-controlled randomized clinical trial that recruited 12,000 women in The Gambia and Burkina Faso to receive either oral azithromycin (2g) or placebo during labor (ratio 1:1).Women, 16+ years were consented during ante-natal visits and enrolled in the trial after oral consent at study health facilities during labor [18].(www.clinicaltrials.govNCT03199547).

Study sites
In The Gambia, women were recruited from two peri-urban government health facilities located close to the capital, Banjul.In Burkina Faso, women were recruited in eight health facilities in rural central districts of Nanoro and Yako (Supplementary Figure 1).The carriage sub-study A sub-group of 250 mother/infant pairs per country participated in this sub-study.They were enrolled into the trial between 23 rd January 2019 and 27 th March 2020 in The Gambia, and between 02 nd April 2019 and 08 th April 2020 in Burkina Faso.
Biological samples were collected pre-intervention until 4 months post-intervention.A maternal nasopharyngeal (NPS) and rectovaginal swab (RVS) were collected during labor before the intervention.Within 4 hours after birth, an NPS and a Rectal Swab (RS) were collected from newborns.Additional samples were collected during household visits: from mothers; NPS at day-6, breast milk (BM) at day-6, 28 and month-4 and from infants; NPS and RS at day-6, 28 and month-4.For The Gambia, the last two sample collection timepoints were affected by the state of emergency declared in March 2020 due to the COVID-19 pandemic [19].

Sample handling and laboratory methods
RVS were collected using a sterile cotton swab inserted 2-3 cm into the vagina, rotated in circular motion for 5 seconds.The same swab was inserted 2-3 cm through the anal sphincter, rotated in circular motion for 5 seconds.The latter procedure was done to collect RS from infants.Sample collection for NPS and BM samples was done as previously described [20].Swabs were placed in a vial containing skim milk-tryptone-glucose-glycerol transport medium in a cold box and transported to the labs within 8 hours.On arrival samples were vortexed for 20 seconds and stored at -70°C for batch processing.Samples collected in Burkina Faso were shipped to The Gambia on dry ice and stored as described above.

Identification of E. Coli and K. Pneumoniae
E. coli and K. pneumoniae were isolated from mothers' BM and RVS and newborns' RS.In addition, K. pneumoniae was isolated from participants' NPS (Supplementary Figure 2).Samples thawed on ice, were vortexed briefly and an aliquot of 50µl dispensed onto MacConkey agar (Oxoid, UK) was streaked for selective isolation of E. coli and K. pneumoniae as previously described [16].For E. coli, identification was done for each morphologically distinct suspected colony when more than one was available and each stored separately.

Statistical analysis
The prevalence of bacterial carriage and antibiotic resistance was compared between trial arms using PR, with corresponding 95% CI's.Fisher's exact test was used to obtain p-values, p<0.05 was considered significant.Stata version 18 was used for all analyses.In the main analyses, the total number of available samples was used as denominator.When we identified >1 E. coli isolate per RVS and RS, to calculate prevalence of E. coli carriage at a particular site we considered a participant a "carrier" if at least one E. coli was isolated from the sample, and a sample was considered "resistant" for a specific antibiotic if at least one resistant E. coli isolate was present.For K. pneumoniae, only one isolate per sample was identified and tested for resistance.In addition, we determined the frequency of antibiotic resistance in infants' RS for samples that were positive for E. coli and K. pneumoniae.

Ethical considerations
The trial was approved by the joint Gambia Government/Medical Research Council Unit The Gambia (MRCG) Ethics committee, the Comité d'Ethique pour la Recherche en Santé (CERS) of Burkina Faso and the LSHTM Ethics Committee.Consent was sought concurrently for both the main trial and carriage sub-study when pregnant women attended antenatal clinics.

RESULTS
Overall, 500 mother-infant pairs participated in this sub-study, 250 from The Gambia, 250 from Burkina Faso (122 in azithromycin and 128 in placebo arm per country).The proportion of samples collected was >98% at day-0 and 6, 92% at day-28, and 79% at month-4.Details of samples available at each timepoint are in Figure 1 (Trial profile).Baseline characteristics of study arms are shown in Table 1.

Prevalence of E. Coli resistance to other antibiotics
Study women.For pre-intervention RVS and post-intervention BM, there were no differences between arms in the prevalence of E. coli resistant to ampicillin, trimethoprim-sulfamethoxazole, gentamicin, and ciprofloxacin.For pre-intervention RVS, there was higher prevalence of ESBL carriage (2.0% vs. 0% p=0.027) in azithromycin arm compared to placebo (Supplementary Tables 3a/3b/3c).There was no resistance to meropenem and cefoxitin resistance was low.

Prevalence of K. Pneumoniae resistance to other antibiotics
Study women.For pre-intervention maternal RVS and post-intervention BM there were no differences between arms in the prevalence of K. pneumoniae resistant to trimethoprimsulfamethoxazole, gentamicin, ciprofloxacin, and ESBL carriage.For pre-intervention maternal RVS, prevalence of K. pneumoniae resistant to cefoxitin was higher in the azithromycin arm (2.0% vs. 0%, p=0.027) (Supplementary Table 6c).No resistance to meropenem was detected.For maternal NPS, resistance to all antibiotics was either absent or low.

DISCUSSION
Clinical trials have shown that prophylactic intrapartum azithromycin decreases maternal and neonatal infections [12,13].It is important, therefore, to evaluate the effect of this intervention on bacterial colonization and antimicrobial resistance.Previous studies showed the intervention decreases carriage of the main gram-positive bacteria causing sepsis in mothers and newborns, with very little effect on azithromycin resistance [6,15].In this study, azithromycin reduced E. coli carriage and increased K. pneumoniae carriage, predominantly in infants' RS.The intervention increased the carriage of azithromycin-resistant isolates for both bacteria.It simultaneously decreased the carriage of E. coli resistant to other antibiotics and increased the carriage of K. pneumoniae resistant to other antibiotics.
In a previous trial conducted in The Gambia following the same design, azithromycin (2g), remained in the maternal breast milk for at least four weeks post-intervention, reaching peak levels on day-6 [24].The substantial concentration of azithromycin transferred to infants, coupled with the impact on maternal carriage, likely explains the effects observed in infants in this study.However, such an effect on RS carriage of E. coli only lasted the neonatal period.These findings are consistent with the effect of azithromycin mass drug administration that reduced the short-term risk of diarrhea in infants aged 2-59 months; diarrheagenic E. coli being a major cause of diarrhea at this age [25][26][27].In vitro experiments have also shown azithromycin is efficacious against certain strains of pathogenic E. coli [28].Moreover, azithromycin can effectively reduce bacterial shedding in patients with shiga-toxin-producing enteroaggregative E. coli (STEC) and travellers' diarrhoea caused by enterotoxigenic E. coli (ETEC) [29,30].Azithromycin use, particularly in mass drug administration campaigns, increased carriage of azithromycin-resistant E. coli [31][32][33], as observed for RS in our study.We observed similar results in a previous study using vaginal samples collected 8-10 days after the intervention [16].The observation that at 4 months post-intervention, the carriage of azithromycin-resistant E. coli is similar between study arms, suggests that the effect on azithromycin resistance is probably waning with decreased drug pressure.
The increased azithromycin resistance was not matched by an increased E. coli resistance to other antibiotics.On the contrary, infants whose mothers had taken intrapartum azithromycin had a lower prevalence of ampicillin, trimethoprim-sulfamethoxazole, and cefoxitin-resistant E. coli isolates.There are two plausible explanations; the lower use of prescribed antibiotics in infants from the azithromycin arm due to lower rates of infections observed during the trial [12] may have resulted in a lower selective pressure and thus lower resistance to common antibiotics.Ampicillin and trimethoprim-sulfamethoxazole are broad-spectrum antibiotics often used for the treatment of respiratory, gastrointestinal and urinary tract infections in West Africa [34,35].Secondly, lower overall prevalence of E. coli carriage in RS would translate to a lower prevalence of carriage of isolates resistant to other antibiotics.The similar frequency of E. coli isolates resistant to the different antibiotics in both arms would support this last hypothesis.This decreased prevalence of carriage of E. coli resistant isolates is an encouraging result that needs to be interpreted considering further evaluation of the overall effects on the microbiome and resistome by the intervention.
We previously showed a higher carriage of K. pneumoniae isolated in BM samples collected after azithromycin treatment [16].In this study, intrapartum azithromycin increased the risk of K. pneumoniae carriage in infants' RS.The strong effect of azithromycin on gram-positive bacteria [6,15] and certain gram-negative bacteria as observed with E. coli here, may have advantaged K. pneumoniae at these body sites.Overgrowth of certain bacterial species after using broadspectrum antibiotics has been reported.A study investigating the effect of early-life antibiotics on the developing infant gut showed that antibiotic-treated infants had a higher abundance of Klebsiella spp.[36].Nevertheless, this higher K. pneumoniae carriage did not increase the incidence of K. pneumoniae sepsis in our PregnAnZI-2 trial [12] or the A-PLUS trial (conducted in 7 low-and middle-income countries) [13].
In our study, it is possible that the higher prevalence of K. pneumoniae carriage resulted in a high carriage of azithromycin-resistant strains in RS, as we previously showed for BM [16].Indeed, the time of the highest carriage of K. pneumoniae (day-28), coincides with that of the highest prevalence of azithromycin-resistant K. pneumoniae isolates in infants' RS.In addition, day-28 was also the timepoint with higher resistance to other tested antibiotics, possibly caused by the same phenomenon.In support of this, we observed a similar trend in the frequency of resistant isolates for all antibiotics, including azithromycin.The production of ESBL in Enterobacterales mediates simultaneous acquisition of resistance to other classes of antibiotics because resistance genes may be located on the same mobile genetic elements [37] and could have contributed to the increased resistance to other antibiotics in K. pneumoniae at day-28.
This study had some limitations.Although we have shown the effect of the intervention on resistance to azithromycin and other antibiotics in E. coli and K. pneumoniae, we could not ascertain the mechanisms of resistance involved.This requires genomic evaluation to complement phenotypic observations.Also, despite the reduction of E. coli carriage following intrapartum azithromycin, it was not possible to determine whether such reduction is beneficial as we have not distinguished between pathogenic and non-pathogenic E. coli.

CONCLUSION
Intrapartum azithromycin decreases carriage of E. coli and increases carriage K. pneumoniae in the gut of neonates.The intervention also increases carriage of azithromycin-resistant E. coli and K. pneumoniae isolates, a potential threat to the spread of such resistance to the community.Conversely, this intervention may decrease resistance to other commonly used antibiotics such as ampicillin or trimethoprim-sulfamethoxazole in E. coli either because it decreases carriage or antibiotic prescription.These results need to be considered when evaluating the overall impact of the use of azithromycin to prevent maternal, neonatal, or infant infections.

Figure 1
Figure 1: Study profile

Figure 2 :
Figure 2: RS E. coli and K. pneumoniae carriage and azithromycin resistance

Table 1 :
Baseline characteristics a Age missing in n = 52 b Season of delivery missing in n = 3 c Birth Weight missing in n = 1.

Table 2 :
Prevalence of E. coli carriage and azithromycin resistance in different biological samples from women and their infants

Table 3 :
Prevalence of K. pneumoniae carriage and azithromycin resistance in different biological samples from women and their infants Downloaded from https://academic.oup.com/cid/advance-article/doi/10.1093/cid/ciae280/7674988 by guest on 29 May 2024 Antibiotic concentration: a azithromycin (0.016-256µg/mL), 1 Samples collected at Day 0, pre-intervention 2 Samples collected at Day 0, post-intervention Isolates with MICs ≥32 µg/mL considered resistant based on azithromycin epidemiological cutoff values and limited clinical data for other Enterobacterales p-values from Fisher's exact test Downloaded from https://academic.oup.com/cid/advance-article/doi/10.1093/cid/ciae280/7674988 by guest on 29 May 2024 a