Molecular Testing for Gastrointestinal Pathogens in Intestinal Tissue of Infants with Necrotizing Enterocolitis or Spontaneous Intestinal Perforation

Abstract


INTRODUCTION
Necrotizing enterocolitis (NEC) and spontaneous intestinal perforation (SIP) are major causes of gastrointestinal morbidity and mortality in preterm infants 1,2 .While NEC is associated with intestinal dysbiosis 3,4 , formula feeding 5 , intestinal ischemia 6 , and prematurity 7 , SIP is a distinctly different entity that primarily affects extremely premature infants during the rst week of age 2 .The histopathological hallmark of NEC is mucosal necrosis, whereas SIP is characterized by mucosal hyperplasia with an absence of the muscularis layer that leads to intestinal perforation 8 .The mechanisms of injury in both conditions remain poorly understood, but underlying in ammation has been implicated in their pathogenesis.
Microorganisms colonizing the preterm gastrointestinal tract may contribute to the pathogenesis of NEC and SIP 9 .Studies of infant stool have revealed dysbiotic signatures of the bacterial microbiome before the onset of NEC, while preclinical models of NEC have shown the absence of disease in germ-free animals 10,11 .Pneumatosis intestinalis, the radiographic and pathologic hallmark of NEC, is believed to occur secondary to gas formation from Gramnegative bacteria, speci cally Clostridioides spp., within the intestinal wall 12 .In addition, NEC cases in neonatal intensive care units (NICUs) can occur in clusters 13 or may be associated with seasonal viral outbreaks 14,15 .
The use of molecular technologies such as polymerase chain reaction (PCR) has facilitated the detection of intestinal pathogens causing communityacquired gastroenteritis in children and adults, yet this technology has not been applied to the evaluation of preterm infants with NEC or SIP.Therefore, the objective of this study was to determine the frequency of typical gastrointestinal bacterial, parasitic, and viral pathogens in intestinal tissue of premature infants with NEC or SIP in order to understand their potential impact on disease pathogenesis 16,17 .

Study Population:
This was a retrospective cohort study of infants with surgical NEC (Bell's ≥ 2B) 18 or SIP who were treated in the Level 4 NICU at Nationwide Children's Hospital (NCH), Columbus, OH from 2000 to 2016.The study leveraged existing specimens from a previously published study examining the detection of cytomegalovirus (CMV) in intestinal tissue of preterm infants with NEC or SIP 19 .As in our prior study, cases were identi ed by review of the surgical pathology database of the Pathology Department, using the diagnoses of "necrotizing enterocolitis" or "small bowel perforation."NEC or SIP was con rmed after histopathologic review by a pediatric pathologist.Inclusion criteria were: (i) histopathological-con rmed diagnosis of NEC or SIP; and (ii) su cient para n-embedded tissue available in the pathology archive.Exclusion criteria were: (i) histopathologic diagnosis of non-NEC or SIP gastrointestinal disease (e.g.atresia, volvulus, omphalocele, gastroschisis); (ii) presence of congenital heart disease; (iii) no retrievable specimens in the pathology archives; or (iv) non-NICU patients.
The medical records of infants were reviewed for pertinent demographic, clinical, laboratory, and radiographic data, and clinical outcomes including retinopathy of prematurity, bronchopulmonary dysplasia (BPD), and short bowel syndrome.The study was approved by the NCH Institutional Review Board (IRB#15-00553).
Tissue Processing and FilmArray® Gastrointestinal Panel (GIFA) (Bio re® Diagnostics, Salt Lake City, UT) Testing: Table 1 lists the gastrointestinal pathogens included in the GIFA panel 16 .Para n-embedded intestinal tissues were processed and DNA/RNA were extracted as previously described 19 .For each patient sample, 10 µl of DNA and 20 µl of RNA were combined with 700 µl of sample buffer and analyzed off-label using the multiplex GIFA PCR panel (BioFire® Diagnostics, St. Lake City Utah).

Statistical Analyses:
Descriptive analyses were used to summarize patients' demographic characteristics using means with standard deviation or medians (interquartile ranges) and frequency distributions as appropriate.Categorical variables were analyzed using Chi-square or Fisher's exact tests, and continuous variables using Ttest or Mann-Whitney tests according to data distribution.Correlations were performed using Spearman's rank correlation coe cient since most of the data did not follow a normal distribution.All analyses were performed using Prism 9.0 (GraphPad Software, LaJolla CA).Two-sided p-values < 0.05 were considered statistically signi cant.

RESULTS
From 2000 to 2016, of 310 infants identi ed in the Pathology database with a diagnosis of "necrotizing enterocolitis" or "small bowel perforation," 178 (57%) cases had either NEC (n = 143; 80%) or SIP (n = 35; 20%) based upon intraoperative surgical reports, histopathologic con rmation of the diagnosis by pathologist review, and availability of formalin-xed, para n-embedded tissue (Fig. 1).Of the 143 infants with NEC, intestinal tissue samples had been obtained at the time of surgery for NEC (86%, n = 123) or at autopsy (14%, n = 20).Among the 35 infants with SIP, 32 (91%) intestinal tissue samples had been obtained at surgery while 3 (9%) were from autopsy.The GIFA detected pathogens in 6 (4%) of the 143 NEC cases including 4 from surgical and 2 from autopsy specimens (Fig. 1).The detected GI pathogens included Clostridioides di cile (C.di cile, n = 2), Enteroaggregative E.coli (EAEC, n = 3), and Giardia lamblia (G.lamblia, n = 1).No viral gastrointestinal pathogens include in the GIFA were detected in NEC or SIP cases.No pathogens were detected by GIFA in intestinal tissue of SIP cases.Assay validation after stool nucleic acid extraction was con rmed with positive GIFA tests for clinically identi ed pathogens in all positive control stool specimens and negative tests for all negative control stools.

Characteristics of GIFA-positive NEC cases:
Gestational ages of GIFA-positive patients ranged from 25 to 34 weeks and birth weight ranged from 715 grams to 2597 grams, and the age of onset of NEC ranging from 5 days to 23 days postnatal age (Table 2).There was no difference in breastmilk vs. formula feeding among the GIFA-positive cases.One case (EAEC) was delivered to a mother with chorioamnionitis.We compared the clinical and laboratory characteristics of GIFA-positive NEC cases and GIFA-negative NEC/SIP patients (Table 3).There were no differences among the GIFA-positive and GIFA-negative groups in demographic characteristics of gender, race or ethnicity, nor were there differences in gestational age or birth weight (Table 3).There was no difference in any breastmilk feedings between the GIFA-positive and -negative groups (67% vs. 59%, p = 0.999).We then investigated clinical signs and symptoms at the time of NEC or SIP presentation that included (i) acute abdominal changes including distention, tenderness or absent bowel sound, (ii) bloody stools, (iii) stability on room air or need for respiratory support, and (iv) gastric residuals.There were no differences in presenting symptomatology between both groups.Analysis of speci c laboratory ndings (complete blood count and liver function tests) revealed no differences in complete blood count at the time of NEC or SIP presentation among each group.However, there was signi cantly higher (p = 0.027) direct bilirubin in infants with GIFA-negative NEC or SIP compared to those with GIFA-positive testing (Table 3).Major sequelae from NEC diagnosis in GIFA-positive cases included death (n = 3, 50%), bronchopulmonary dysplasia (BPD, n = 1) and short bowel syndrome (SBS, n = 1) (Table 3).Perinatal Characteristics of GIFA-positive cases: All mothers of the GIFA-positive patients were multiparous with 2 gravid status (Table 2).There was no signi cant difference among the different types of pathogens detected and the mode of birth delivery (50% vaginal delivery, 50% caesarean section).

DISCUSSION
The contribution of intestinal pathogens causing community-acquired gastroenteritis to the pathogenesis of NEC or SIP remains unknown.In this study, using a commercially available multiplex PCR-based assay, we detected C. di cile, Enteraggregative E.coli (EAEC), and G. lamblia in 6 of 143 infants with NEC Stage ≥ 2B but no pathogens in 35 SIP cases.Although we previously detected CMV by PCR or immunohistochemistry (IHC) in the same intestinal tissue samples biorepository 19 , CMV is not included in the GIFA panel and no viral gastrointestinal pathogens were detected using GIFA.A study by Ullrich et al similarly investigated the presence of gastrointestinal pathogens in 22 NEC ileal samples compared to 15 non-NEC controls using a different multiplex-PCR panel 20 .This study investigated the same common gastrointestinal pathogens as our study, however, results did not show the presence of any viral, bacterial or protozoan pathogens in NEC intestinal tissue 20 .Our larger subject cohort and, perhaps, differences in test sensitivity of the GIFA could have contributed to the difference in our results.Our ndings indicate that typical GI pathogens are unlikely to be primary causes of NEC, but may support a rationale for further investigation into the role of gastrointestinal pathogens in in ammation during NEC.Our results also indicate that microbes are unlikely to contribute to SIP pathogenesis.
The colonization of the preterm gut with different Clostridioides species was rst described by Ferraris et al with the most common including (1) C. perfringens, (2) C. butyricum, (3) C. di cile, and (4) C. paraputri cum 21 .Studies have shown that the NICU environment and antibiotic exposures were major in uencing exposures in preterm neonates 22 .Several Clostridioides species, C. neonatale 23 , C. di cil 24 , and C. perfringens 25 , isolated from blood, stool, and peritoneal samples have been associated with NEC outbreaks.One longitudinal study described the preterm gut microbiome in those who developed NEC compared to controls as temporally distinguished by the abundance of Clostridioides compared to E.coli.Stool samples from infants with early onset NEC (de ned as < 14 days of age) were characterized by an abundance of Clostridioides species (mainly C. sensu stricto) vs. those who developed late-onset NEC with an increased abundance of Gammaproteobacteria (E.coli and Shigella) 26 .Similarly, our C. di cile-associated NEC cases occurred at day of life (DOL) 10 which could be classi ed as early onset NEC compared to the majority of EAEC-associated NEC cases that occurred at DOL > 20.
The association of E.coli subtype colonization and NEC have been described in several studies.A case report described of the presence of E.coli O157:H7 in a term infant who developed NEC 27 .Preclinical animal models of NEC have demonstrated opposing effects of E.coli subtypes on disease severity.Thomas et al describes how the colonization of the commensal strain E. coli EC25 protected against experimental NEC 28 , while Roy et al demonstrated in a higher level preclinical model in piglets that E.coli-fermented short chain fatty acid metabolites in formula induced a more severe form of NEC that mimicked the human form of the disease 29 .Other studies have characterized the preterm gut microbiota as less diverse in those that develop NEC 30 with a potential in uence of uropathogenic E. coli as a risk factor for increased severity of NEC and death 31 .
To date, there are no reported cases of G. lamblia associated NEC cases in the literature.However, Giardia infection has been shown to alter the human bacterial microbiome structure and function inducing a dysbiotic environment after the enteropathogenic organism has been cleared 32 .Although our study did not detect common viral gastrointestinal pathogens (e.g.rotavirus, norovirus, astrovirus), a recent meta-analysis did nd a signi cant association between CMV, rotavirus, norovirus, and astrovirus infection and increased risk for NEC 33 .
Limitations of our study include a single center study with retrospectively collected samples from requiring surgical intervention or with fatal outcome.The GIFA panel is indicated for stool samples and limits the pathogen detection to those included in the panel.Its use on formalin-xed tissue is not included in the FDA label and has undetermined performance characteristics.Furthermore, the processing of formalin-xed tissue may have impacted the sensitivity of pathogen detection 34 .Finally, our study methodology cannot distinguish colonization from causation, limiting conclusions of microbial contributions.
In summary, the detection of C. di cile, E.coli subtype-EAEC and G. lamblia in intestinal tissues from cases of severe NEC may provide insight into the role of common gastrointestinal pathogens as possible infections contributors to intestinal in ammation and necrosis with progression to surgical NEC in preterm infants.These associations may deserve further study among diverse clinical populations and preclinical animal models.

Declarations
Con ict of Interest: Authors have no con icts of interest or competing nancial interests to report.
Data Availability: The data that supports the ndings in this study are available from the last author (MS), upon request.
Funding Source: This work was supported in part by Nationwide Children's Hospital intramural grant to DTM and PJS (Grant #20066515).The sponsors had no role in study design, data collection, data analysis, data interpretation, decision to publish, or preparation of the manuscript.
Author contributions: designed the study, data analysis, drafted and revised the nal manuscript.PJS: designed the study, obtained funding for the study, reviewed and revised the nal manuscript.MC: processed the data, reviewed and revised the nal manuscript.IK: processed the data, reviewed and revised the nal manuscript.KE: responsible for data analysis, data processing, reviewed and revised the nal manuscript.AM: responsible for data analysis, data processing, reviewed and revised the nal manuscript.DM: designed the study, obtained funding of the study, reviewed and revised the nal manuscript.MM: designed the study, reviewed and revised the nal manuscript.SPG: data acquisition, data processing, reviewed and revised the nal manuscript.MS: designed the study, data analysis, drafted and revised the nal manuscript.All authors approved the nal manuscript as submitted and agree to be accountable for all aspects of the work.Cases of NEC or SIP were classi ed by outcome (surgical or autopsy)."Other" consisted of ileal strictures, omphalocele, and abdominal hernia.

Figure 1 Schematic
Figure 1Schematic description of study population.The study population included 310 total cases from 2000-2016 with "necrotizing enterocolitis" or "small bowel perforation" diagnosis in the pathology database.Cases (132 samples) were excluded due to congenital anomalies, non-NICU patients, and those with no medical record or tissue available.The study population included 178 (57%) of infants with histopathologically con rmed NEC (n=143) and SIP (n=35).

Table 1
, except for omission of the para n lysis step.For the extracted control nucleic acids, 10 µl of DNA and 20 µl of RNA were mixed with 700µl of sample buffer and tested using the GIFA panel to validate pathogen detection using this nucleic acid extraction method.

Table 3
Clinical and laboratory characteristics of GIFA-positive and GIFA-negative NEC/SIP patients.