“Enhanced acquisition of antibiotic-resistant intestinal E. coli during the first year of life assessed in a prospective cohort study”

Escherichia coli (E. coli) colonize the gut of newborns within hours after birth when the newborn is exposed to the maternal vaginal and fecal flora and to environmental bacteria [1, 2]. Under physiological circumstances E. coli living in the colonic mucus and its human host are symbionts [3]. Nevertheless, virulent E. coli strains can severely sicken neonates and infants, causing enterocolitis, urinary tract infections, meningitis, and septicemia [4‐8].

Increased antibiotic resistance in E. coli has been observed in recent years [9‐11]; antibiotic-resistant pathogens are generally a growing menace [12‐16], especially in low-income countries, where prevalence of resistant bacteria is higher than in high-income countries [17]. In Thailand, rates of antibiotic resistance in E. coli are high in healthy adults, food, food animals, and the environment [18].

In literature, there are only few data on the prevalence of antibiotic-resistant bacteria in healthy newborns [19, 20]. Smit and coworkers have found a high rate of neonates (46%, n = 17) to be colonized with multidrug-resistant Klebsiella pneumoniae within 24 h of admission to a neonatal unit in Cambodia and they have claimed a high within-host diversity [21]. It is well known that antibiotic exposure triggers the development of resistant bacteria [22]. However, it is still a matter of debate how infants acquire resistant bacteria including commensal bacteria of the intestinal microbiota without any antibiotic selective pressure. We sought to investigate the acquisition of antibiotic resistance in E. coli from stool of healthy infants in Northern Thailand during the first year of life and to assess whether resistant E. coli persist in infant gut as well as whether these microorganisms are transmitted from parent to child. We also investigated which host and environmental factors and which bacterial virulence factors had contributed to frequency and duration of infant gut colonization by antibiotic-resistant E. coli.

This cohort study was approved, according to the declaration of Helsinki 2000, by the ethics committees of the Medical University of Innsbruck (Ethics approval number: 260/2010) and the Faculty of Medicine, Chiang Mai University (Ethics approval number: 260/2533). Samples were collected between September 2010 and December 2012 at the Department of Pediatrics, Chiang Mai University, and the Meledek (Mother and Child) Hospital, both Chiang Mai, and at the Sanpatong District Hospital, all in Northern Thailand after informed consent of the parents was obtained. Characterizations of the resistant strains were completed in 2017.

During the enrollment period 437 babies were born in the participating hospitals (Fig. 1). Among the latter, 169 newborns and their parents fulfilled the inclusion criteria and consented to participate in the study. At the first sampling time point (within 48 h after birth) 169 meconium, 169 maternal and 164 paternal stool samples were collected. During the whole study period 27 families were culled due to lack of more than one stool sample, thus 142 families completed the study (Fig. 1). Among these 142 participants stool samples from 4 infants were not received at 2wk and from other 8 infants at 4 to 6mo.

Ninety of the 142 infants were born at Chiang Mai University Hospital (63,4%), 31 at Chiang Mai Meledek Hospital (urban child birth clinic) (21,8%), and 21 at the rural Sanpatong District Hospital (14,8%); 75 were male (52.8%) and 67 female (47.2%).

All infants were born between week 32 and week 42 of gestation, with 20 (14.1%) before week 37 and 122 (85.9%) were term births. In 103 cases vaginal birth was spontaneous and in 7 it required vacuum support; 32 babies were born by cesarean section (Table 1).

A culture-based approach was chosen due to several advantages compared to other methods like sequence- based tools especially the possibility to characterize single bacterial species in a stool sample. This method is unique for a big sample size as in the present study and a long observation period of 1 year, however, it might have limitations which are discussed below.

In particular we found high proportions of resistance to TET, AM, and SXT, ranging from 20 to 30%, in E. coli from meconium of neonates in northern Thailand. The prevalence of antibiotic-resistant E. coli increased during the first year of life, reaching almost the same proportions as in parental samples, with > 70% resistance to TET and AM and 66% to SXT. Resistance against CEF was less frequent, with 8% in meconium samples and 33% in parental stool samples. In 8 children (6%) resistant E. coli strains persisted over at least three sampling time points. Early transmission of resistant E. coli from parents to children was identified in 8 families (6%). In neither persistent nor transmitted strains an association with any tested virulence and adhesion factor was observed.

The neonatal phase is crucial for intestinal colonization with E. coli [1, 28, 29], although this may occur even during the fetal period [30‐32].

Of our 142 meconium samples, 47 yielded resistant E. coli (33%). Compared with other studies this rate of antibiotic resistance is high: In Innsbruck, Austria, among 21 neonates studied for resistant E. coli during the first 48 h of life, resistant E. coli (resistant to TET, AM, and SXT) was observed in only one stool sample (4.8%) [19]. AM- resistant E. coli were found in 27 of 824 (3.3%) neonatal stool samples in Lithuania [20]. The differences in resistance prevalence may be inferred from differences in study design and geographical background depended on the study location.

Countries belonging to the Organization for Economic Co-operation and Development (OECD) are generally wealthy. In asymptomatic infants antibiotic resistance is much higher in non-OECD countries than in OECD countries: The pooled resistance prevalence in OECD countries to TET was 37.7% (25.9 to 49.7%), to AM, 37.6% (24.9 to 54.3%), and to SXT, 33.5% (2.2 to 71.0%). Resistance in non-OECD countries was higher, with resistance to TET 80.0% (59.7 to 95.3%); to AM, 67.2% (45.8 to 84.9%); and to SXT, 61.4% (no CI given) [17].

Our study confirmed that resistance proportions are high in Thailand, a non-OECD country, with resistance proportions of 80% to TET, 72% to AM, and 66% to SXT in one-year old infants.

One reason for the high prevalence of antibiotic resistance in Thailand in healthy children may lie in wide use of antibiotics, which are available at grocery stores and retail shops without prescription [33]. Moreover, high proportions of antibiotic resistance among bacteria have been found in various environmental sources in Thailand, including farms, water, and food [18]. Additionally, extensive use of antibiotics in livestock worsens the situation by transmission of resistant strains to humans via the food chain [34, 35].

The prevalence of carriage of third-generation cephalosporine-resistant E. coli in the present study (both among infants and parents) was rather low compared with other regional investigations [18, 36‐39]. Luvsansharav et al. showed local differences in prevalence of ESBL- producing E. coli in healthy individuals in different provinces in Thailand varying between 32 and 53.9% [36]. In the present study resistance against the third-generation cephalosporine ceftriaxone was found in only 20%. We did not perform a selective screening for third-generation cephalosporine-resistant E. coli using ESBL-chromogenic media or MacConkey agar supplemented with a third-generation cephalosporine. This might have resulted in a low sensitivity.

Both antibiotic use and day-care center exposure have been described as risk factors for the acquisition of resistant E. coli strains in neonates and young children [40, 41]. We found no statistically significant association between antibiotic use and occurrence of resistant E. coli.

However, that most children who received antibiotic therapy did so aged between 6mo and 1 yr., and that at 6mo the prevalence of antibiotic resistance was already above 80%, makes it difficult to draw conclusions.

In the present study no data on the attendance of day-care centers were collected. However, attendance of day care centers is common in Thailand and thus it can be assumed that this also had an influence on transmission and colonization with resistant E.coli in the present study.

Several factors have been proposed as influencing the composition of the intestinal microbiota of infants. In the KOALA Birth Cohort Study of 1032 infants aged 1mo in the Netherlands, mode of delivery, type of infant feeding (breast feeding versus formula feeding), gestational age, and antibiotic use were the most important determinants for development of intestinal flora [42]. In the present study we also found significant lower prevalence of resistant E. coli in meconium in children born by caesarian section compared to vaginal delivery. This is consistent with the findings of previous studies showing that delivery mode affects the direct transmission of initial bacteria from mother to newborns [43, 44]. In contrast, birth in the rural hospital was associated with an increased risk of early colonization with resistant E. coli strains, antibiotic usage by the mother during pregnancy showed to be protective. We can only speculate about the reasons for this. Less hygienic standards in rural hospitals might be an explanation for this. Some studies showed placental transfer of antibiotics such as erythromycin in humans [45]. This might be a reason for delayed colonization with intestinal resistant bacteria in neonates. However, the reason why maternal antibiotic usage was protective in early colonization with resistant E. coli in the present study is not clear and needs further evaluation.

We did not find that feeding behavior [46] or antibiotic usage of the child affected gut colonization with resistant E. coli. Other determinants associated with prevalence of resistant E. coli in meconium like siblings or pets, were also not identified as significant in our study.

We are aware that our study has several limitations. First, parents were taught by nurses how to collect and to store stool samples and how to keep the samples cooled during transport to the laboratory. Nevertheless, mistakes in sample handling by parents cannot be excluded.

Furthermore, sample collection was initially planned at ages 4mo and 6mo. A flood in northern Thailand in 2011 isolated some study participants for several weeks. The 4mo and 6mo sample data therefore were pooled.

Third, bacterial strains can be excreted intermittently. Thus by taking only one stool sample per time point we cannot exclude to have missed resistant E. coli. Furthermore, we cannot exclude that detection of a certain strain was influenced by our sampling processing protocol (including the colony picking) or by any unknown factors. Thus, our conclusions on transmission and persistence have to be interpreted with caution. Moreover, antibiotic use in infants during the first year of life may have influenced the diagnostic sensitivity of the culture method.

Another main limitation is the limited number of study participants which decreases the power of our study and thus the ability to detect further determinants associated with development of resistance.