Gut Pathogen Colonization: A Risk Factor to Bloodstream Infections in Preterm Neonates Admitted in the Neonatal Intensive Care Unit – A Prospective Cohort Study

In neonatal intensive care units (NICUs) across the globe, neonates, particularly preterm neonates, face heightened and unique risk of infections due to their immature immune system, barrier function, and peristalsis [1, 2]. According to the Global Antibiotic Research & Development Partnership (GARDP), bloodstream infection (BSI) is defined as presence of viable microorganisms in the bloodstream that elicit or have elicited an inflammatory response characterized by the alteration of clinical, hemodynamic, and laboratory parameters [3]. A systematic analysis evaluating trends in child mortality for a period of 10 years revealed that BSIs are responsible for 13% of total neonatal mortality within the first week of life [4].

The gastrointestinal tract of neonates is a complex ecosystem hosting billions of microorganisms that play pivotal roles in early immune development. However, the gut can also act as reservoirs for numerous pathogens [5, 6]. Emerging evidence suggests that colonization of gut by certain pathogens is not mere coincidence but may be a precursor to more severe systemic infections. This phenomenon, known as gut pathogen colonization, has garnered significant attention in recent years, especially in the context of its potential to precipitate BSIs in neonates in the NICU. Colonization resistance is a process whereby the normal gut microbiota resists the invasion of exogenous pathogens and expands resident pathobionts [7, 8]. Disruption of colonization resistance leads to overgrowth of opportunistic bacterial species, which are usually present in low numbers but are capable of multiplying to high levels under disrupted conditions [9, 10]. This overgrowth of microbes can harm the host and paves the way for pathogenic bacteria, including members of the Enterobacteriaceae family, to capitalize on weakened gut colonization, enabling colonization of the gut [11, 12].

This disruption allows bacteria to translocate into the bloodstream and ultimately results in BSIs. Pathogens responsible for BSIs are frequently present in the environment of the NICU and are commonly found in the gut microbiome of hospitalized neonates [9, 13]. Hence, we hypothesize that gut pathogen colonization is a significant precursor to the development of BSIs in neonates admitted to the NICU. This study investigates the association between gut pathogen colonization, hypothesizing that the species causing BSI primarily originated from the gut.

A prospective controlled cohort study was conducted in the NICU of a tertiary care hospital in Karnataka, India. This study received approval from the Institutional Ethics Committee (IEC: 490/2020) and was registered prospectively with the Clinical Trials Registry – India (CTRI/2020/11/029375). From January 2021 to September 2023, inborn preterm infants (gestational age <37 weeks) with birth weights <2,500 g were recruited.

Sample size calculation was based on the difference in the proportion of Klebsiella spp. colonization derived from a pilot study. The pilot study consisted of 22 samples, with 11 in the case group and 11 in the control group, revealing colonization proportions of 40% and 20%, respectively. We aimed to detect this clinically significant difference of 20% with 80% power and a 5% level of significance calculated using the formula for “Sample Size Determination in Comparing Two Proportions.” This yielded a sample size of approximately 82 per group. To account for potential attrition or contamination, we increased this by 10%, resulting in 91 samples per group.

All the eligible inborn preterm infants admitted to NICU were recruited at birth (day 1) and followed prospectively, observing them over time to identify the occurrence of sepsis. The neonates were followed prospectively for 28 days for onset of sepsis. After the follow-up period, we divided the infants into groups based on whether they developed sepsis. Sepsis was defined as all neonates who developed late-onset sepsis during the study period. Demographic details of recruited neonates, such as gender, gestational age, birth weight, and morbidities, were collected from case records. For group A, the first stool sample was taken on day 4 of life, and the second sample was taken on day 14. For group B, the first stool sample was collected on the 4th day of life (stool culture 1), and the second sample was obtained on the day when the Gram stain of blood culture bottles showed growth of pathogenic bacteria, confirming sepsis (stool culture 2). Stool samples were aseptically collected using a sterile spoon from the diapers of the neonates and transferred into sterile containers. These samples were transported within 4 h to a microbiology laboratory for conventional culture workup, which included both aerobic and anaerobic cultures. Gut-blood microbial correlation in this study is defined as the concurrence of similar bacteria in the gastrointestinal tract as well as in the bloodstream.

The diagnostic culture workup involved inoculating specimens on culture media to target both aerobic and anaerobic bacteria. Stool samples were inoculated onto 5% sheep blood agar and MacConkey agar plates. These plates were incubated at 37°C and examined for bacterial growth after 24 h and 48 h. Bacterial isolates were identified using Matrix-assisted laser desorption/ionization-time of flight mass spectrometry (MALDI-TOF) (Vitek MS, bioMerieux Inc., France). The stool sample was divided into two parts for anaerobic culture. One part was inoculated onto 5% sheep blood agar and neomycin blood agar with a metronidazole disc (5 U), then incubated at 37°C for 72 h in a Whitley A35 Anaerobic Workstation (Don Whitley Scientific, Shipley, UK). The other part of the stool was inoculated into Robertson’s cooked meat medium and incubated at 37°C. Gram staining was performed on the 3rd, 5th, and 7th day of inoculation, with subculturing onto 5% sheep blood agar if any new morphology was observed. Bacterial isolates were identified using MALDI-TOF MS.

Patient demographics and bacterial isolate susceptibility patterns were reported as percentages, means (±SD), and medians (±IQR). Means (±SD) were used to summarize data for continuous variables with a normal distribution, while medians and ranges were used for data with a skewed distribution. Linear regression analysis was used to predict the gut-blood correlation effect by regressor variables.

A total of 453 neonates were initially screened for eligibility, of which 183 were excluded for various reasons. After additional exclusions during the 28-day follow-up period, 159 neonates were included in the final analysis, divided into group A (no sepsis, n = 91) and group B (sepsis, n = 68). The detailed participant flow, including specific reasons for exclusion at each stage, is presented in Figure 1. In group B, the average age at sepsis diagnosis was 10 ± 6 days, and the number of days of antibiotics prior to day 14 was 6 ± 4 days (Table 1). The mean age at diagnosis was 9.5 ± 6 days for Gram-negative bacteria and 8 ± 5 days for Gram-positive bacteria. Klebsiella pneumoniae was the most common pathogen, causing 44.11% (n = 30), followed by Acinetobacter baumannii 16.2% (n = 11).

The analysis of gut microbiome colonization revealed a notably higher abundance of potential pathogens in the neonates with sepsis compared to the non-septic group. For group A, on day 4, K. pneumoniae was detected in 38.5% (n = 35) of the stool samples, E. coli in 33.0% % (n = 30), and Bifidobacterium spp. 49.5% (n = 45). By day 14, the colonization rates for K. pneumoniae and E. coli both increased to 49.5% and 46.2%, respectively, while Bifidobacterium spp. increased to 68.1%. In contrast, group B exhibited different colonization patterns. For stool culture 1, K. pneumoniae was present in 67.6% of the samples, E. coli in 60.3%, and Enterococcus faecalis in 27.9%. For stool culture 2, the colonization of K. pneumoniae and E. coli decreased to 58.8% and 44.1%, respectively, while E. faecalis and Bifidobacterium spp. increased to 29.4% and 32.4%, respectively (Fig. 2). These findings suggest a strong correlation between the high abundance of pathogenic bacteria in the gut and the subsequent development of BSIs in preterm neonates.

We investigated the association between the presence of pathogen microbiota in the gut that was also present in the blood – defined as gut-blood microbial presence – and its association with the mode of delivery, along with the presence of various bacteria in neonates. Table 2 presents the results of a multiple regression analysis conducted to identify predictors influencing the co-occurrence of gut and bloodstream pathogens in preterm neonates for stool culture 1 (Fig. 3). The analysis reveals that cesarean delivery was a statistically significant predictor, with a negative coefficient of −0.244 (p = 0.050), indicating that neonates delivered by cesarean section had a reduced likelihood of gut-blood bacterial co-occurrence compared to those delivered vaginally. Another significant predictor was the result of blood cultures, with a positive coefficient of 0.105 (p = 0.001), suggesting that positive blood cultures strongly correlate with the presence of the same pathogens in the gut. Although K. pneumoniae colonization showed a positive association (B = 0.325), this result did not reach statistical significance (p = 0.232). Overall, the model accounted for 30.4% of variance (R² = 0.304, F = 3.215, p = 0.004), indicating a moderate relationship between the selected predictors and the co-occurrence of bacterial colonization in both gut and bloodstream.

The regression model for stool culture 2 further highlights the significant role of gut pathogen colonization in the co-occurrence of BSI. K. pneumoniae remained a significant predictor of bacterial co-occurrence between the gut and bloodstream, with a coefficient of 0.517 (p = 0.017). This emphasizes the ongoing risk of K. pneumoniae colonization in the development of BSIs as the neonates aged. Additionally, the blood culture results continued to be a strong predictor (B = 0.138, p < 0.001), reinforcing the relationship between positive blood cultures and the presence of identical pathogens in the gut, which mirrors the findings from stool culture 1 (Table 3). Overall, the model explained 42.0% of the variance (R² = 0.420), indicating a moderate-to-strong relationship between the predictors and the co-occurrence of bacterial colonization in both the gut and bloodstream. The F statistic of 4.052 (p = 0.001) further confirms the model’s significance. These findings, in continuation with the stool culture 1 results, suggest that persistent colonization of K. pneumoniae in the gut is a critical risk factor for the development of BSIs in preterm neonates, as colonization patterns evolve over time.

The relationship between gut pathogen colonization and BSIs in preterm neonates admitted to NICUs is a subject of critical importance because the immature intestinal barrier can facilitate the translocation of pathogens into the bloodstream [14, 15]. The presence of pathogens such as K. pneumoniae and E. coli (both members of the family Enterobacteriaceae and phylum Proteobacteria) in the gut prior to the onset of sepsis in our study aligns with findings from existing literature [16], suggesting a potential pathway for these microbes to translocate from gut to the bloodstream. A significant proportion of neonates had the bloodstream pathogen present in gut before the clinical onset of sepsis, suggesting a direct link between gut colonization and subsequent BSIs.

Schwartz et al. [17] highlighted the significance of a balanced gut microbiome in preventing pathogen proliferation and infections. The study also revealed that there was a significantly greater abundance of the causative species from the Enterobacteriaceae families in the gut than in healthy neonates. Furthermore, the presence of Proteobacteria in sepsis-affected neonates in our study is consistent with findings from other studies, such as the study by Lu et al. [18], which reported a high abundance of this phylum in relation to neonatal sepsis. This shift in microbial populations could be indicative of an environment more conducive to pathogen proliferation and might play a role in the pathogenesis of sepsis in these infants [19, 20]. Recent evidence suggests that central line days are linked to an increased risk of BSI in NICUs, with prolonged catheter use poses a significant risk for CLABSI in infants [21]. A study in low- and middle-income countries reported a pooled CLABSI rate of 4.82 per 1,000 catheter days, higher than in high-income countries [22]. Probiotics also show promise in reducing BSI, showing an 84% reduction in infection rates [23]. However, due to the lack of data on which neonates received probiotics and had central lines, this study was not able to perform an analysis on these variables.

In a study by Lee et al. [24], which explored the link between gut dysbiosis and neonatal sepsis, the study concluded that the composition of the gut microbiome in preterm infants was similar to that in healthy infants at birth but evolved toward dysbiosis with increasing Proteobacteria and decreasing Firmicutes weeks later [24]. Schwartz et al. [17] in his research revealed that 58% of gut microbiomes before BSI and 79% (15 out of 19) of gut microbiomes at any time contained the BSI isolate, showing fewer than 20 genomic substitutions. In contrast, our study focused on patients with BSI, where each infant provided a stool sample before BSI onset. In these samples, the causative species made up more than 10% of the gut microbiota. Remarkably, in 75% of these cases, the stool samples exhibited over 45% of the BSI causing species, indicating a significant presence of the causative bacteria in the gut prior to the infection. In conclusion, this study provides valuable insights into the complex interplay between the gut microbiome and BSIs in neonates in NICU settings.

This study has several strengths, including its prospective controlled cohort design, which provides a robust framework for establishing associations between gut pathogen colonization and BSIs in preterm neonates. However, there are limitations to consider. The study is conducted at a single tertiary care hospital in Karnataka, India, which may limit the generalizability of the findings to other settings. Furthermore, while the study establishes an association between gut colonization and BSIs, it cannot definitively prove causation. Additionally, the removal of infants with early-onset sepsis reduced the sample size, preventing the study from reaching its original target. Despite this limitation, the remaining sample provided a valuable insight into associations between gut pathogen colonization and BSIs in preterm. The absence of sequencing data further limits the depth of microbial analysis, as more comprehensive sequencing methods would provide a clearer understanding of the gut microbiota composition. Another limitation of this study is lack of data on probiotic administration and central line use in enrolled neonates as were unable to evaluate their direct impact. Future research should incorporate detailed tracking of these factors to better understand their role in neonatal sepsis.

The study revealed that there is a significant relationship between gut pathogen colonization and the emergence of BSIs in preterm neonates admitted to the NICU. The study highlights that certain pathogen, such as K. pneumoniae and E.coli, are present in the gut prior to sepsis onset. Compared with their healthier counterparts, neonates with sepsis exhibit distinct gut microbiome compositions. This conclusion points to the potential of using gut microbiota modulation as a preventive strategy against BSIs in vulnerable infants and emphasizes the need for ongoing research to further understand and manage sepsis in preterm neonates.

The authors would like to thank Kasturba Medical College, Manipal Academy of Higher Education, for providing technical support.

This study protocol was reviewed and approved by the Institutional Ethics Committee, Kasturba Medical College and Kasturba Hospital, approval No. IEC:490/2020. Written informed consent was obtained from either of the parent of recruited neonate.

The authors have no conflicts of interest to declare.

This study was not supported by any sponsor or funder.

F.I. performed the experiments, analyzed and interpreted the data, and contributed reagents, materials, analysis tools, or data; F.I. and N.S. wrote the manuscript and analyzed and interpreted the data; J.P., N.S., and L.E.S.L. contributed reagents, materials, analysis tools, or data; P.A.S. and V.K.E. analyzed and interpreted the data.

All data generated or analyzed during this study are included in this article. Further inquiries can be directed to the corresponding author.