WO2014152985A1 - Biomarkers of preterm necrotizing enterocolitis - Google Patents

Abstract

Methods for identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC) based on metabolomics biomarkers in urine samples and/or microbial biomarkers in stool samples. Preterm birth contributes disproportionately to the global burden of morbidity and mortality in infancy. Necrotizing enterocolitis (NEC) is a devastating emergency of preterm infants that affects about 10% of infants born <29 weeks gestational age, with a case-fatality of about 30%.

Description

BIOMARKERS OF PRETERM NECROTIZING ENTEROCOLITIS

Cross-Reference to Related Applications

This application claims the benefit of the filing date of U.S. Provisional Application No. 61/783,018, filed March 14, 2013, the entire contents of which are incorporated by reference herein.

Government Support

This invention was made with government support under Grant Number HD059140 awarded by the U.S. National Institutes of Health (National Institute of Child Health and Human Development). The government has certain rights in the invention.

Background of the Invention

Preterm birth contributes disproportionately to the global burden of morbidity and mortality in infancy. Necrotizing enterocolitis (NEC) is a devastating emergency of preterm infants that affects about 10% of infants born <29 weeks gestational age, with a case-fatality

1 2

of about 30%/' " NEC typically occurs without clinical warning between 3 days and several months of postnatal life.³ Risk factors for NEC include immaturity,⁴ timing and type of infant feeding,⁵' ⁶ extended empirical use of antibiotics,⁷' ⁸ and intestinal bacterial

9-12

colonization. ^" Unfortunately, after decades of research, the specific pathogenesis of NEC remains an enigma, and validated biomarkers for identifying individuals at highest risk of disease are lacking. Summary of the Invention

The present disclosure is based, in part, on the unexpected discoveries that necrotizing enterocolitis (NEC) in infants is preceded by distinct metabolic and/or microbial signatures. For example, alanine, vitamin B₆, and vitamin B₆ metabolite were found to be positively associated with NEC in infants; and histidine was found to be inversely associated with NEC in infants. It was also discovered that infants with an increase of alanine relative to histidine in the first 9 days after birth are at risk for subsequent development of NEC. As another example, Firmicutes dysbiosis (e.g. , imbalance of Firmicutes bacteria) was found to be positively associated with NEC in infants. The present disclosure is also based on the unexpected discoveries that there is a relationship between the metabolic signatures and the microbial signatures associated with NEC in infants. For example, alanine was found to be positively associated with NEC cases that are preceded by Firmicutes dysbiosis, and histidine was found to be inversely associated with NEC cases that not preceded by high Firmicutes dysbiosis or NEC cases that are preceded by high Proteobacteria. It was also discovered that a high alanine:histidine ratio predicts NEC onset and correlates with microbial characteristics that predict NEC. It was also discovered that a combination of high relative abundance of Firmicutes in the microbiota of a stool sample, a high ratio of alanine to histidine in a urine sample, and an increased urinary glucose concentation, as compared to matched control levels, are predictive of NEC. Such discoveries are particularly useful in, for example, diagnosing and/or assessing risk of developing NEC in preterm infants.

Accordingly, provided here are methods for identifying preterm infants having or at risk for NEC (e.g. , subtypes of NEC) based on any of the metabolic biomarkers and microbial biomarkers described herein, or a combination thereof. A preterm infant identified in any of the methods described herein as having or at risk for NEC can be subjected to a suitable treatment for NEC as known in the art, including those described herein.

In one aspect, the present disclosure relate to methods of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the methods comprising providing a urine sample of a preterm infant, measuring a level of a metabolite biomarker in the urine sample, wherein the metabolite biomarker is vitamin B₆ (also known as pyridoxine) a vitamin B₆ metabolite (e.g. , 4-pyridoxate), alanine, histidine, tyrosine, or a combination thereof, and identifying the preterm infant as having or at risk for NEC if the level of the metabolite biomarker deviates from a reference level.

In some embodiments, the urine sample is obtained from the preterm infant during postnatal day 4 to postnatal day 9.

In some embodiments, the metabolite biomarker is vitamin B₆ or a vitamin B₆ metabolite. The preterm infant is identified as an infant having or at risk for type I NEC if the level of the vitamin B₆ or the vitamin B₆ metabolite in the urine sample is elevated relative to the reference level. In some embodiments, the metabolite biomarker is alanine. The preterm infant is identified as an infant having or at risk for type I NEC if the level of alanine in the urine sample is elevated relative to the reference level.

In some embodiments, the metabolite biomarker is a combination of alanine and histidine. The method can comprise measuring the levels of histidine and alanine in the urine sample, calculating a ratio between the level of alanine and the level of histidine, and identifying the infant as having or at risk for NEC if the ratio deviates from a reference level. For example, the infant is identified as having or at risk for NEC if the ratio of the level of alanine to the level of histidine is greater than 4.

In some embodiments, the metabolite biomarker is histidine. The preterm infant is identified as an infant having or at risk for type II NEC if the level of histidine is reduced relative to the reference level.

Alternatively, the metabolite biomarker can be tyrosine. The preterm infant is identified as an infant having or at risk for type II NEC if the level of tyrosine is reduced relative to the reference level.

In any of the methods described above, the level of the metabolite biomarker can be determined by nuclear magnetic resonance (NMR) spectroscopy.

In another aspect, the present disclosure provides a method of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the method comprising: providing at least one stool sample of the preterm infant; determining bacteria diversity of the microbiota in the stool sample; and identifying the preterm infant as an infant having or at risk for developing NEC based on the diversity of the microbiota.

The at least one stool sample can be obtained from the preterm infant during postnatal day 4 to postnatal day 9. In some instances, the preterm infant is identified as an infant having or at risk for type I NEC if the diversity of the microbiota represents a high relative abundance of Firmicutes (e.g., equal to or greater than 80%, 90%, 95%, or 98%). In one example, the Firmicutes are Bacilli bacteria. In another example, the Firmicutes are

Staphylococcaceae bacteria (e.g., Staphylococcus bacteria), Enterococcaceae bacteria (e.g., Enterococcus bacteria), or both.

In some instances, the preterm infant is identified as an infant having or at risk for type I NEC if the diversity of the microbiota represents a high relative abundance of

Firmicutes (e.g., equal to or greater than 80%, 90%, 95%, or 98%). In one example, the Firmicutes are Bacilli bacteria. In another example, the Firmicutes are Staphylococcaceae bacteria (e.g., Staphylococcus bacteria), Enterococcaceae bacteria (e.g., Enterococcus bacteria), or both.

Alternatively or in addition, the at least one stool sample can be obtained during postnatal day 10 to postnatal day 16. In some instances, the preterm infant is identified as an infant having or at risk for type II NEC if the bacterial diversity of the microbiota represents a high relative abundance of Proteobacteria (e.g., equal to or greater than 80%, 90%, 95%, or 98%). In one example, the Proteobacteria are Enterobacteriaceae bacteria (e.g., Escherichia bacteria, Enterobacter bacteria, or both).

Alternatively or in addition, the at least one stool sample can be obtained during postnatal day 1 to postnatal day 9 (e.g., postnatal day 1 to postnatal 7 or postnatal day 4 to postnatal day 9). In some instances, the preterm infant is identified as an infant having or at risk for type IIB NEC if the bacterial diversity of the microbiota represents a high relative abundance of Proteobacteria {e.g., equal to or greater than 80%, 90%, 95%, or 98%). In one example, the Proteobacteria are Enterobacteriaceae bacteria {e.g., Escherichia bacteria, Enterobacter bacteria, or both).

Alternatively or in addition, the at least one stool sample can be obtained during postnatal day 8 to postnatal day 21. In some instances, the preterm infant is identified as an infant having or at risk for type I or IIB NEC if the bacterial diversity of the microbiota represents a high relative abundance of Proteobacteria (e.g., equal to or greater than 80%, 90%, 95%, or 98%). In one example, the Proteobacteria are Enterobacteriaceae bacteria (e.g., Escherichia bacteria, Enterobacter bacteria, or both).

In some embodiments, the at least one stool sample includes a first stool sample and a second stool sample, the first stool sample being obtained from the preterm infant during postnatal day 4 to postnatal day 9 and the second stool sample being obtained from the preterm infant during postnatal day 10 to postnatal day 16. The preterm infant can be identified as an infant having or at risk for NEC if the bacteria diversity of the microbiota in the second stool sample represents a decrease in the relative abundance of Firmicutes, an increase in the relative abundance of Proteobacteria, or both, as compared to the bacteria diversity of the microbiota in the first stool sample. The Firmicutes can be Bacilli bacteria or Staphylococcaceae bacteria (e.g., Staphylococcus bacteria), Enterococcaceae bacteria {e.g., Enterococcus bacteria), or both. Alternatively or in addition, the Proteobacteria are

Enterobacteriaceae bacteria (e.g., Escherichia bacteria, Enterobacter bacteria, or both).

In any of the methods described herein that involves a microbial biomarker, the bacterial diversity in a biosample of interest (e.g., a urine or stool sample) can be determined by analyzing the 16s RNAs amplified from the sample.

In yet another aspect, the present disclosure provides a method of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the method comprising: providing at least one stool sample and at least one urine sample of the preterm infant;

determinining bacteria diversity of the microbiota in the stool sample; measuring a level of a metabolite biomarker in the urine sample, wherein the metabolite biomarker is vitamin B₆, a vitamin B₆ metabolite, alanine, histidine, or a combination thereof; and identifying the preterm infant as an infant having or at risk for developing NEC based on the bacteria diversity and the level of the metabolite.

In some embodiments, both the urine sample and the stool sample are obtained during postnatal day 4 to postnatal day 9. In one example, the metabolite is alanine, vitamin B₆, a vitamin B₆ metabolite, or a combination thereof. The preterm infant is identified as an infant having or at risk for type I NEC if the level of the metabolite is higher than a reference level and the bacteria diversity represents a high relative abundance of Firmicutes.

In other embodiments, the urine sample is obtained during postnatal day 4 to postnatal day 9 and the stool sample is obtained during postnatal day 10 to postnatal day 16. In one example, the metabolite is histidine. The preterm infant can be identified as an infant having or at risk for type II NEC if the level of histidine is lower than a reference level and the bacteria diversity represents a high relative abundance of Proteobacteria.

If desired, the level of the metabolite can be measured by NMR. Alternatively or in addition, the bacteria diversity is determined by analyzing the 16s RNAs amplified from the stool sample.

In another one aspect, the present disclosure relate to methods of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the methods comprising (a) providing a urine sample of a preterm infant, (b) measuring a level of a metabolite marker, which can be alanine, histidine, glucose, or a combination thereof, in the urine sample, (c) providing a stool sample of the preterm infact, (d) determining bacteria diversity of the microbiota in the stool sample , and (e) identifying the preterm infant as having or at risk for NEC based on the level of the metabolite marker in the urine sample and the diversity of the microbiota in the stool sample. In some examples, the method involves calculating a ratio between the level of alanine and the level of histidine. In some embodiments, whether the preterm infant has or is at risk for NEC is determined based on (i) the alanine/ histidine ratio in the urine sample, (ii) the glucose concentration in the urine sample, and (iii) the diversity of the microbiota in the stool sample. For example, the preterm infant can be identified as having or at risk for NEC, if the alanine/ histidine ratio deviates a reference level (e.g., greater than 4), the glucose concentration is higher than a reference level, and the diversity of the microbiota represents a high relative abundance of Firmicutes (e.g., equal to or greater than 80%, 90%, 95%, or 98%). In one example, the Firmicutes are Bacilli bacteria. In another example, the Firmicutes are Staphylococcaceae bacteria (e.g., Staphylococcus bacteria), Enterococcaceae bacteria (e.g., Enterococcus bacteria), or a combination thereof. The level of metabolite marker and the bacterial diversity may be measured using any method known in the art, including those described herein (e.g., by NMR spectroscopy and by 16s RNA analysis, respectively). In some embodiments, the NEC is NEC without prior or concurrent sepsis.

Any of the methods described herein can further comprise subjecting the preterm infant to a treatment of NEC if the preterm infant is identified as having or at risk for NEC.

Also within the scope of the present disclosure are kits for use in identifying preterm infants who have or are at risk for NEC, each of the kit comprising one or more reagents for determining the level of one or more of the metabolic and/or microbial biomarkers as described herein (e.g., nucleotide probes/primers for measuring and/or amplifying 16s RNAs from a suitable biosample such as a urine sample or stool sample) and optionally instructions for using the kit.

The details of one or more embodiments of the invention are set forth in the description below. Other features or advantages of the present invention will be apparent from the following drawings and detailed description of several embodiments, and also from the appended claims. Brief Description of the Drawings

In the drawings:

FIGURE 1A shows the relative abundance of bacterial phyla in infants who developed NEC versus control infants. Columns represent samples from days of life 4-9 and 10-16. Data are graphed as box plots. The middle bar represents the median, the outer horizontal lines of the box represent the 25th and 75th percentiles, and vertical lines represent proximal values. The dots overlaying the plots indicate the values of individual samples.

FIGURE IB shows box plots of the phylogenetic distribution of the intestinal microbial communities of case and control infants at the level of the bacterial phyla (first column) and genera (second column). Boxes indicate the 75th and 25th percentiles, with the median indicated by the bar midway across the box. Vertical bars indicate proximal values. The number of samples (and infants) included in each time window and case-control group are indicated in the upper right corner of the graphs. The phyla are color-coded. The color- coding for the genera follows indicates the phylum under which they are classified (for example, the relative abundance of genus Entewbacter, represented in teal, indicates that it is of the phylum Proteobacteria). In this study population, the preponderance of the microbial community composition consisted of Proteobacteria and Firmicutes. The relative bacterial abundance among study infants is indicated in decreasing order from left to right.

FIGURE 2 shows microbial community differences between NEC and control infants, days 4 to 9 samples (Panel A). This box plot indicates the relative abundance of the genus Propionibacterium in 18 control samples and 9 samples prior to NEC onset during days of life 4 to 9. None of the infants who later developed NEC had detectable amounts of

Propionibacterium, but 10 (56%) of the control samples had detectable amounts of

Propionibacterium (Fisher's exact, p=0.009). Panel B: This box plot indicates the Chaol richness index in samples from days of life 4 to 9. NEC cases tended towards lower diversity than controls, but the comparison was not significant (Kruskal-Wallis, p=0.086).

FIGURE 3 shows non-metric multidimensional scaling (NMDS) ordination of microbial communities for days of life 4-9 (Panel A). This ordination was based on weighted UniFrac beta-diversity and run with 3 dimensions in the vegan package of R, resulting in a stress of 4.06. Control samples are shown in black and NEC and non-NEC deaths in either red (for samples included in Cluster- 1) or green (for samples included in Cluster- II). Clusters of samples with similar microbial composition were systematically identified using the Ward minimum variance method. These clusters are labeled using roman numerals I through IV. All NEC cases were found in either Cluster-I or Cluster- II only. The samples of the two non- NEC deaths are also found as part of Cluster-I. Panel B: Bars indicate the relative abundance of the 10 most common genera in samples from individual infants, whose study numbers are noted on the horizontal axis. Samples are grouped according to case or control status and the cluster in which they were identified.

FIGURE 4 shows NMDS ordination of microbial communities for days of life 10-16 (Panel A). This ordination was based on weighted UniFrac beta-diversity and run with 3 dimensions in the vegan package of R, resulting in a stress of 2.43. Controls are shown in black and NEC and non-NEC deaths in either red (cases identified as NEC-I in the days 4 to 9 ordination) or green (cases identified as NEC-II in the days 4 to 9 ordination). Clusters identified using the Ward minimum variance method are indicated in this ordination as A, B, and D; C is identified as an outlier value. Panel B: Bars indicate the relative abundance of the 10 most common genera in samples from individual infants; study numbers are noted on the horizontal axis. NEC sub-types (NEC-I and NEC-II) correspond to the NEC cases included in Cluster-I and Cluster- II, respectively, in the ordination of days 4 to 9 sample (FIG. 3). The clusters identified in this ordination are indicated by column headers. Clusters indicate microbial community similarity.

FIGURE 5 shows box plots of urinary metabolites in relation to NEC sub-types versus controls. Panel A: Urinary alanine. Panel B: Urinary histidine. Panel C: Ratio of urinary alanine to histidine.

FIGURE 6 shows ordination after rarefaction of samples.

FIGURE 7 shows NMDS ordination by delivery mode and case control status.

FIGURE 8 shows impact of extraction protocol.

FIGURE 9 shows differences between case types 1 (red) and 2 (green) based on

LEfSe. (A & B) Taxonomy of significant differences (p<0.05) between case type 1 and 2 for infant days of life 4-9 (A) and 10-16 (B). (C & D) Histogram of the linear discriminant analysis (LDA) log scores of features that significantly differ between case types 1 and 2. OUT_2 is an OUT (operational taxonomic unit) of Staphylococcus.

FIGURE 10 shows a comparison of characteristics of case types and controls. (A)

Timing of onset of case types 1 and 2. (B) Histograms comparing the distribution of clinical variables: Case types 1 and 2 and controls. (C) Histograms comparing the distribution of urinary metabolite values (alanine, pyridoxine, histidine and tyrosine): Case types 1 and 2 and controls. Significant differences (p < 0.05) were indicated between cases and controls (*); case type 1 or 2 versus all other infants (+); and case type 1 versus case type 2 (^Λ).

FIGURE 11 shows LEfSe-generated cladogram (The Huttenhower Lab) and linear discriminant analysis (LDA) log scores comparing the microbial composition of samples from all 14 cases and 21 control infants, collected during postnatal days of life 4 to 9. Data for samples from postnatal days of life 4 to 9 are not shown, as there were no significant differences between cases and controls in the microbial composition of samples from days of life 10 to 16.

FIGURE 12 shows NMDS ordination of OTU level data for samples from infant days of life 4-9 (panel A) and 10-16 (panel B) based on a weighted UniFrac distance metric after rarefaction. NMDS was run with 3 dimensions in the vegan package of R. The ordination results are similar.

FIGURE 13 shows Chaol and Simpson a-diversity metrics by day of life window and for samples collected from study infants classified as case-type 1, case type 2, and controls. There was no significant differences by study groups.

FIGURE 14 shows extraction protocol in relation to ordination of samples (White dots=standard; black dots=Qiagen kit). The samples from each extraction protocol are similarly distributed and do not cluster.

FIGURE 15 shows ordinations run with NEC and controls; death only infants excluded. The ordinations find the same clustering even without the three non-NEC deaths included.

FIGURE 16 shows ordination and regression analyses demonstrating that alanine is associated with composition of cases but not controls.

FIGURE 17 shows a series of graphs of non-metric multidimensional scaling

(NMDS) ordination of microbial communities in stool samples for week 1, week 2, and week 3 in a follow-up study.FIGURE 18 shows two graphs of the relative abundance of

Enterobacteriaceae and Proteobacteria in stool samples in infants having sepsis preceding NEC.

FIGURE 19 shows a graph of the relative abundance of Firmicutes in stool samples from infacts having or not having NEC. Crosses = median. Line = 85% relative abundance. FIGURE 20 shows two graphs of the relative abundance of Proteobacteria and Enterobacteriaceae in four groups: controls, Type 3, Type 4, and Type 5 NEC. For each group, the bars are, from left to right: week 1, week 2, and week 3.

FIGURE 21 shows the area under ROC curve for high firmicutes as a marker for NEC not preceded by sepsis.

FIGURE 22 shows a graph of the alanine-to-histidine ratio in urine samples from infants having NEC (1) and infants not having NEC (0).

Detailed Description of the Invention

Necrotizing enterocolitis (NEC) is the most common gastrointestinal medical/surgical emergency occurring in infants. It is an acute inflammatory disease characterized by variable damage to the intestinal tract ranging from mucosal injury to full-thickness necrosis and perforation. Early colonizing organisms interact with the intestinal mucosa to shape the developing immune system towards homeostasis or dysregulation, ^19-24 and might thus contribute to the pathobiology leading to onset of NEC. To address this, the early microbial community was examined to identify predictive microbial biomarkers of later NEC in a prospective study of preterm infants. Culture-independent 16S rRNA gene sequencing of stool samples from 4 to 16 days of life was utilized to identify microbial community signatures. Because intestinal bacteria can influence the metabolic profiles of their hosts, 25-29 urinary metabolomics was also pursued to identify surrogate biomarkers of NEC.

Identifying Subjects Having or At Risk for NEC

Accordingly, the present disclosure is based, in part, on the unexpected discoveries of microbial and/or metabolic signatures useful for predicting NEC onset, including early-onset NEC (type I NEC), late-onset NEC (type II NEC), sepsis-NEC (type IIA NEC), and other NECs (type IIB NEC). See Table 1 and Examples below.

Accordingly, described herein are methods for identifying preterm infants at risk for NEC based on any of the metabolic biomarkers, microbial biomarkers, or a combination thereof as described herein. A preterm infant identified as having or at risk for NEC by any of the methods described herein can be subjected to a suitable treatment known in the art, including those described herein.

A "preterm infant," as used herein, may refer to a human baby born before 37 weeks of gestation. For example, a preterm infant may be born before 37, 36, 35, 34, 33, 32, 31, 30, 29, 28, 27, 26, 25 or 24 weeks of gestation. Full term infants are typically about 37 to about 42 weeks of gestation. In some embodiments, a preterm infant may have a birth weight of less than about 2500 grams (g). For example, a preterm infant may have a birth weight of less than about 2400 g, 2300 g, 2200 g, 2100 g, 2000 g, 1900 g, 1800 g, 1700 g, 1600 g, 1500 g, 1400 g, 1300 g, 1200 g, 1100 g, 1000 g, or less. In some embodiments, a preterm infant may be born at less than about 29 weeks of gestation (or at about 29 weeks of gestation) and have a birth weight of less than about 1200 g (or about 1200 g). In some embodiments, the preterm infants are free of congenital anomalies (e.g., malformations, birth defects). In some embodiments, the preterm infants are also free of NEC (or NEC is not detectable) within the first week of postnatal life.

A "preterm infant at risk for necrotizing colitis," as used herein, refers to a preterm infant who is at increased risk of developing NEC, as compared to a matched control who survives free of NEC and/or free of sepsis. Sepsis refers to an illness in which the body has a severe response to bacteria or other germs. A "matched control" is a term of art and herein refers to a preterm infant who is matched by comparable (e.g., the same or closely similar) parameters such as gestational age, birth weight, sex/gender, race, mode of delivery (e.g., Cesarean section), maternal antibiotic used at time of delivery, and infant antibiotic used.

Necrotizing Enterocolitis

Necrotizing enterocolitis (NEC) is a gastrointestinal disease that mostly affects premature infants. NEC involves infection and inflammation that causes destruction of the bowel (intestine) or part of the bowel. NEC usually occurs within the first 2 weeks of life, usually after milk feeding has begun (at first, feedings are usually given through a tube that goes directly to the infant's stomach). About 10% of infants weighing less than 3 lbs. 5 oz. (1,500 grams) experience NEC. These preterm infants have immature bowels, which are sensitive to changes in blood flow and prone to infection. They may have difficulty with blood and oxygen circulation and digestion, which increases their chances of developing

NEC. In some embodiments, NEC may be defined as modified Bell's stage II or III. 30

The present disclosure describes two NEC sub-types, type I (NEC-I), type II (NEC- II), depending upon the timing of disease onset, e.g., days 7-21 after birth versus 19-39 after birth. Each type of NEC is associated with distinct metabolite phenotypes and/or distinct intestinal microbial phenotypes, which precede early or late NEC, or death. The present disclosure also describes two sub-classes of type II NEC, type IIA (NEC-IIA) and type IIB (NEC-IIB). These two sub-classes of type II are each associated with distinct metabolite phenotypes and/or distinct intestinal microbial phenotypes, which precede early or late NEC, or death. A summary of the NEC sub-types is provided in Table 1. Table 1. NEC Sub-types and Sub-Classes

NEC-IIA (subclass A of type Late onset (see above); • high relative abundance of II; type IIA) sepsis occurring prior to phylum Proteobacteria and

or concurrently with family Enterobacteriaceae NEC onset, median

gestational age of 25

weeks, maternal

antibiotics given at the

time of delivery

NEC-IIB (subset B of type Late onset (see above); • high relative abundance of II, type IIB) median gestational age of phylum Proteobacteria and

25 weeks, no evident family Enterobacteriaceae in difference than control stool samples during week 2 samples at week 1 and/or week 3 of life

The term "high" or "low" used in Table 1 refers to the level of a marker relative to a control level of the same marker as described herein. The onset of type I NEC may be between days of life 7 and 21. Infants who develop type I NEC may be characterized by a high level of the metabolite alanine, as compared to matched control levels of alanine, during the first week of life (e.g., days of life 4 to 6). Infants who develop type I NEC may also be characterized by a high level of vitamin B₆ (also referred to as pyridoxine) or a vitamin B₆ metabolite such as, for example, 4-pyridoxate, as compared to matched control levels, during the first week of life (e.g., days of life 4 to 6). In some embodiments, infants who develop type I NEC may have a high level of alanine and a high level of vitamin B₆ (and/or a vitamin B₆ metabolite), as compared to matched controls, during days of life 4 to 6.

Infants who develop type I NEC may also be characterized by a high relative abundance of Firmicutes in the microbiota of a stool sample of the infants during the first week of life (e.g., days of life 4 to 6), referred to herein as Firmicutes dysbiosis. In some embodiments, Firmicutes bacteria comprise at least about 50% median relative abundance of the microbiota of type I NEC infants. For example, Firmicutes may comprise about 75%, about 76%, about 77%, about 78%, about 79%, about 80%, about 81%, about 82%, about 83%, about 84%, about 85%, about 86%, about 87%, about 88%, about 89%, about 90%, about 91%, about 92%, about 93%, about 94%, about 95%, about 96%, about 97%, about 98% or about 99% median relative abundance of the mictobiota of type I NEC infants.

Infants who develop type I NEC may also be characterized by low relative abundance of Proteobacteria in the microbiota. In some embodiments, Proteobacteria comprise less than about 20% or less than about 10% median relative abundance of the mictobiota of type I NEC infants. For example, Proteobacteria may comprise about 20%, about 19%, about 18%, about 17%, about 16%, about 15%, about 14%, about 13%, about 12%, about 11%, about 10%, about 9%, about 8%, about 7%, about 6%, about 5%, about 4%, about 3%, about 2% , about 1%, or less. In some embodiments, infants who develop type I NEC do not have

Propionibacteria present in their microbiota.

Thus, in some embodiments, alanine is positively associated with NEC cases that are preceded by Firmicutes dysbiosis. In some embodiments, alanine and vitamin B₆, or a vitamin B₆ metabolite is positively associated with NEC cases that are preceded by

Firmicutes dysbiosis.

In some embodiments, infants who develop type I NEC may be characterized by a high level of alanine, as compared to matched controls, together with Firmicutes dysbiosis and an absence of Propionibacteria during days of life 4 to 6. In some embodiments, infants who develop type I NEC may be characterized by a high level of alanine and a high level of vitamin B₆, or a vitamin B₆ metabolite, as compared to matched controls, together with Firmicutes dysbiosis and an absence of Propionibacteria during days of life 4 to 6.

Infants having or at risk of developing type I NEC may also be characterized by a trend to C-section delivery and/or median gestational age of 27 weeks. In some

embodiments, infants who develop or are at risk of developing type I NEC may be characterized by a high relative abundance of Firmicutes in the microbiota of a stool sample of the infants, e.g., during the first week of life or during days 4 to 9. In some embodiments, the high relative abundance of Firmicutes in the microbiota of a stool sample of the infants is greater than 80%, 85%, 90%, or 95%. In some embodiments, the high relative abundance of Firmicutes in the microbiota of a stool sample of the infants is greater than 85%. In some embodiments, infants who develop or are or at risk of developing type I NEC may be characterized by high relative abundance of phylum Proteobacteria and family

Enterobacteriaceae in stool samples during week 2 and/or week 3 of life.

The onset of type II NEC may be between days of life 19 and 39. Infants who develop type II NEC may be characterized by a low level of the metabolite alanine, as compared to matched control levels of alanine, during days of life 10 to 16. Infants who develop type II NEC may also be characterized by a low level of histidine and/or a low level of tyrosine, as compared to matched control levels, during days of life 10 to 16. In some embodiments, infants who develop type II NEC may have a low level of alanine and a low level of histidine, as compared to matched control levels, during the first week of life (e.g., days of life 4 to 6). In some embodiments, infants who develop type II NEC may have a low level of alanine and a low level of tyrosine, as compared to matched control levels, during the first week of life (e.g., days of life 4 to 6). In some embodiments, infants who develop type II NEC may have a low level of histidine and a low level of tyrosine, as compared to matched control levels, during days of life 4 to 6.

Infants having type II NEC may also be characterized by a high relative abundance of Proteobacteria in the microbiota of the infants' stool samples during days of life 10 to 16, referred to herein as Proteobacteria dysbiosis. In some embodiments, Proteobacteria comprise at least about 50% median relative abundance of the mictobiota of type II NEC infants. For example, Proteobacteria may comprise about 75%, about 76%, about 77%, about 78%, about 79%, about 80%, about 81%, about 82%, about 83%, about 84%, about 85%, about 86%, about 87%, about 88%, about 89%, about 90%, about 91%, about 92%, about 93%, about 94%, about 95%, about 96%, about 97%, about 98% or about 99% median relative abundance of the mictobiota of type II NEC infants. Infants who develop type II NEC may also be characterized by low relative abundance of Firmicutes. In some embodiments, Firmicutes comprise less than about 20% or less than about 10% median relative abundance of the mictobiota of type II NEC infants. For example, Firmicutes may comprise about 20%, about 19%, about 18%, about 17%, about 16%, about 15%, about 14%, about 13%, about 12%, about 11%, about 10%, about 9%, about 8%, about 7%, about 6%, about 5%, about 4%, about 3%, about 2%, about 1%, or less. In some embodiments, infants who develop type II NEC do not have Propionibacteria present in their microbiota.

Thus, in some embodiments, histidine is inversely associated with NEC cases that are preceded by Proteobacteria dysbiosis. In some embodiments, histidine and alanine are inversely associated with NEC cases that are preceded by Proteobacteria dysbiosis. In some embodiments, histidine, alanine and tyrosine are inversely associated with NEC cases that are preceded by Proteobacteria dysbiosis.

In some embodiments, infants who develop type II NEC may be characterized by a low level of histidine, as compared to matched control levels, together with Proteobacteria dysbiosis during days of life 10 to 16. In some embodiments, infants who develop type II

NEC may be characterized by a low level of alanine and a low level of histidine, as compared to matched control levels, together with Proteobacteria dysbiosis during days of life 10 to 16. In some embodiments, infants who develop type II NEC may be characterized by a low level of alanine, a low level of histidine, and a low level of tyrosine as compared to matched control levels, together with Proteobacteria dysbiosis and an absence of Propionibacteria during days of life 10 to 16.

It was also discovered that a high ratio of alanine to histidine predicts NEC onset in preterm infants, irrespective of sub-type (predictive value 78%, p=0.007). Thus, in some embodiments, a ratio of alanine:histidine is used as a marker to identify subjects having or at risk for NEC. For example, a ratio of alanine:histidine greater than about 4 is positively associated with NEC onset.

As shown in Table 1 above, there are two subclasses of type II NEC, type IIA and type IIB.

Infants having or at risk of developing type IIA NEC may be characterized by sepsis occurring prior to or concurrently with NEC onset. In some embodiments, infants having or at risk of developing type IIA NEC have a median gestational age of 25 weeks. In some embodiments, infants having or at risk of developing type IIA NEC have maternal antibiotics given at the time of delivery. In some embodiments, infants who develop or are at risk of developing type IIA NEC may be characterized by a high relative abundance of phylum Proteobacteria and family Enterobacteriaceae in stool samples. In some embodiments, the high relative abundance of phylum Proteobacteria and family Enterobacteriaceae is during week 1 of life. Infants having NEC not preceeded by sepsis may be characterized by high relative abundance of Firmicutes in the microbiota of a stool sample of the infants.

Infants having or at risk of developing type IIB NEC may be characterized by a median gestational age of 25 weeks and may not have characteristics associated with other subtypes of NEC. In some embodiments, infants who develop type IIB NEC may be characterized by a high relative abundance of phylum Proteobacteria and family

Enterobacteriaceae in stool samples during week 2 and/or week 3 of life.

In some embodiments, infants having any type of NEC may be characterized by a combination of (a) high relative abundance of Firmicutes in the microbiota of a stool sample, (b) a high ratio of alanine to histidine, and (c) an increased urinary glucose concentation, as compared to matched control levels of these markers. In some embodiments, Firmicutes bacteria comprise at least about 50% median relative abundance of the microbiota. For example, Firmicutes may comprise about 75%, about 76%, about 77%, about 78%, about 79%, about 80%, about 81%, about 82%, about 83%, about 84%, about 85%, about 86%, about 87%, about 88%, about 89%, about 90%, about 91%, about 92%, about 93%, about 94%, about 95%, about 96%, about 97%, about 98% or about 99% median relative abundance of the mictobiota. In some embodiments, the ratio of the level of alanine to the level of histidine is greater than about 4. For example, the ratio of the level of alanine to the level of histidine may be greater than 4, 4.5, 5, 5.5, 6, 6.5, 7, 7.5, 8, 8.5, 9, 9.5 or 10. In some embodiments, the glucose concentration is elevated by at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level.

To use the combined markers as described herein for identifying preterm infants having or at risk for NEC, a stool sample and a urine sample can be obtained from a candidate subject. The levels of alanine, histidine, and glucose in the urine sample can be measured using a routine method. When desired, such levels can be normalized against an internal control. A ratio of alanine to histidine can then be calculated accordingly. The alanine/histidine radio and the level of urinary glucose of the infant can then be compared with corresponding control levels as described herein to determine whether the candidate has an elavoated alanine/histidine ratio and an elevated level of glucose. Further, the diversity of microbiota in the stool sample can be measured using routine technology such as 16s rRNA mapping. The populations of bacteria and their relative abundances can be determined accordingly to determine whether the candidate has a high relative abundance of Firmicutes in his or her stool sample. If the candidate is identified as having or at risk for NEC based on the combination of the alanine/histidine ratio, the abundance of Firmicutes, and the uninary glucose level, the candidate can be subject to a suitable treatment to treat NEC, delay the onset of NEC, or reduce the risk of NEC development. Such treatments are well known in the art, including those described herein.

In some embodiments, infants having sepsis or at risk of developing sepsis may be characterized by a high relative abundance of family Enterobacteriaceae in the microbiota of a stool sample of the infant. In some embodiments, the high relative abundance of family Enterobacteriaceae in the microbiota of a stool sample of the infants is greater than 80%, 85%, 90%, or 95%. In some embodiments, the high relative abundance of family

Enterobacteriaceae in the microbiota of a stool sample of the infants is greater than 87%. Metabolite Biomarkers

A preterm infant may be identified as at risk for NEC based on a relative level of at least one metabolite biomarker. In some instances, the preterm infant may be identified as at risk for type I NEC or type II NEC based on a relative level of at least one metabolite biomarker. In some embodiments, a preterm infant may be identified as at risk for NEC (or type I or type II NEC) based on a relative level of one, two, three or four metabolite biomarkers. A preterm infant is considered to be at risk for NEC if the relative level of at least one metabolite biomarker deviates from a reference level. In some instances, the reference level is obtained from a matched control. A metabolite biomarker level is considered to "deviate" from a reference level if the difference between the reference level and the metabolite level is at least 20%. For example, a metabolite biomarker level indicative of risk of NEC may deviate from a reference level by at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, at least 55%, at least 60%, at least 65%, or more. For example, a metabolite biomarker level that is elevated relative to the reference level by at least 20% may be used to identify a preterm infant as at risk for NEC. Conversely, a metabolite biomarker level that is reduced relative to the reference level by at least 20% may be used to identify a preterm infant as at risk for NEC, depending on the metabolite. In some instances, the level of a metabolite biomarker may distinguish between risk of developing type I NEC and type II NEC. For example, a level of a metabolite biomarker that is elevated relative to a reference level may identify a preterm infant at risk for type I NEC and a level of the same metabolite biomarker that is reduced relative to the reference level may identify the preterm infant as at risk for type II NEC. Likewise, a level of a metabolite biomarker that is reduced relative to a reference level may identify a preterm infant at risk for type I NEC and a level of the same metabolite biomarker that is elevated relative to the reference level may identify the preterm infant as at risk for type II NEC. When two or more metabolite levels are used to identify risk of NEC, at least one metabolite biomarker level may be elevated relative to reference level and at least another metabolite biomarker level may be reduced relative to a different reference level.

(i) Alanine

In some embodiments, the metabolite biomarker is alanine. Alanine (abbreviated as

Ala or A) is an a-amino acid with the chemical formula CH₃CH(NH₂)COOH. A preterm infant may be identified as at risk of NEC if the level of alanine (e.g., obtained from a urine sample of the preterm infant) deviates from a reference level. In some embodiments, a preterm infant is identified as at risk for type I necrotizing enterocolitis if the level of alanine is elevated relative to the reference level. The level of alanine may be elevated by at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level. In some

embodiments, a preterm infant is identified as at risk for type II necrotizing enterocolitis if the level of alanine is reduced relative to the reference level. The level of alanine may be reduced by at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level. (ii) Histidine

In some embodiments, the metabolite biomarker is histidine. Histidine (abbreviated as His or H) is an a- amino acid with an imidazole functional group. A preterm infant may be identified as at risk of NEC if the level of histidine (e.g., obtained from a urine sample of the preterm infant) deviates from a reference level. In some embodiments, a preterm infant is identified as at risk for type II necrotizing enterocolitis if the level of histidine is reduced relative to the reference level. The level of histidine may be reduced by at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level.

In some embodiments, a preterm infant may be identified as at risk of NEC if ratio of the level of alanine to the level of histidine is greater than about 4 (or if the ratio of the level of histidine to the level of alanine is less than about 4). For example, the ratio of the level of alanine to the level of histidine may be greater than 4, 4.5, 5, 5.5, 6, 6.5, 7, 7.5, 8, 8.5, 9, 9.5 or 10. (iii) Vitamin

In some embodiments, the metabolite biomarker is vitamin B₆ or a vitamin B₆ metabolite such as, for example, 4-pyridoxate. Vitamin B₆, also referred to as pyridoxine, is a water-soluble vitamin and is part of the vitamin B complex group. A preterm infant may be identified as at risk of type I NEC if the level of vitamin B₆ or vitamin B₆ metabolite (e.g., obtained from a urine sample of the preterm infant) is elevated relative to the reference level. The level of vitamin B₆ or vitamin B₆ metabolite may be elevated by at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level. (iv) Tyrosine

In some embodiments, the metabolite biomarker is tyrosine. Tyrosine (abbreviated as Tyr or Y) or 4-hydroxyphenylalanine, is a non-essential amino acid with a polar side group. A preterm infant may be identified as at risk of NEC if the level of tyrosine (e.g., obtained from a urine sample of the preterm infant) deviates from a reference level. In some embodiments, a preterm infant is identified as at risk for type II necrotizing enterocolitis if the level of tyrosine is reduced relative to the reference level. The level of tyrosine may be reduced by at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level. (v) Glucose

In some embodiments, the metabolite biomarker is glucose. A preterm infant may be identified as at risk of NEC if the level of glucose (e.g., obtained from a urine sample of the preterm infant) is elevated relative to the reference level. The level of glucose may be elevated by at least 5%, at least 10%, at least 15%, at least 20%, at least 25%, at least 30%, at least 35%, at least 40%, at least 45%, at least 50%, or more relative to the reference level.

The level of a metabolite biomarker may be obtained from a bodily fluid or tissue sample of the preterm infant. For example, the level of a metabolite biomarker may be obtained from a urine sample. Alternatively, the level of a metabolite biomarker may be obtained from a blood sample, for example, from a sample of blood plasma or a sample of blood serum. The bodily fluid or tissue sample may be collected prior to diagnosis of NEC. The sample may be obtained from the preterm infant during postnatal day 4 to postnatal day 9. For example, the sample (e.g. , urine sample) may be obtained on postnatal day 4, postnatal day 5, postnatal day 6, postnatal day 7, postnatal day 8 or postnatal day 9. More than one sample may be collected and analyzed. The samples may be collected on the same day or on different days.

The level of a metabolite biomarker in accordance with the present disclosure may be a nucleic acid level or a protein level. Thus, in some embodiments, the methods provided herein include measuring the level of a nucleic acid (e.g., DNA or RNA), while in other embodiments, the methods include measuring the level of protein. The nucleic acid level or protein level of metabolite biomarker may be determined by any means known to one of ordinary skill in the art. For example, assays for measuring protein levels (e.g.,

concentrations) include, without limitation, absorbance assays (e.g., absorbance at 280 nm or 205 nm, extinction coefficient) and colorimetric assays (e.g., modified Lowry, biuret, Bradford assay, bicinchoninic acid assay (Smith), amido black method, colloidal gold). In some embodiments, nuclear magnetic resonance (NMR) may be used to determine the level of metabolite biomarker.

Microbiota

A preterm infant may be identified as at risk for NEC based on relative abundance of microbiota. "Microbiota," as used herein, refers to the population of microorganisms that typically inhabit a bodily organ or part. Microbiota may also be referred to as flora. The bacteria diversity of the microbiota refers to the types of one or more bacteria found in the microbiota and their relative abundances in the whole population. The bacteria diversity, particularly the relative abundance of one or more particular microorganisms, may be used to identify a preterm infant as at risk of NEC, including type I NEC, type II NEC, type IIA NEC, and/or type IIB NEC. The bacteria diversity in a microbiota from a biosample can be determined via routine methods. For example, alpha-diversity can be examined using the Simpson diversity index and/or Chao 1 richness index (see Chao et al., 1984 and Simpson, 1949). Results from this study indicate that infants who later developed NEC tended towards lower alpha-diversity and lacked Propionibacterium as compared to controls. Thus, low alpha-diversity and/or lacking Propionibacterium could be used as biomarkers for predicting NEC onset.

The dominant phyla of the microbiota of preterm infants are Firmicutes (e.g., Gram- positive organisms) and Proteobacteria (e.g., Gram-negative organisms). The most common genera in preterm infants include, for example, Enterobacter, Staphylococcus, Escherichia, Enterococcus, Leuconostoc, Lactococcus, Streptococcus and Clostridia. Enterobacter, Staphylococcus, Escherichia and Enterococcus account for more than 90% of the microbiota. Control infants typically have a median relative abundance of about 80% Proteobacteria and about 20% Firmicutes, with a small proportion of Bacteriodetes and Actinobacteria.

In some embodiments, a preterm infant may be identified as at risk for NEC (e.g., NEC-I) if the relative abundance of the microbiota represents an elevated level of Firmicutes, as compared to the levels of Firmicutes represented by a reference relative abundance of microbiota. The Firmicutes are a phylum of bacteria, most of which have Gram-positive cell wall structure. In some embodiments, the Firmicutes belong to class Bacilli (e.g., order Bacillales or Lactobacillales). Bacteria belonging to class Bacilli, order Bacillales, include Bacillis, Listeria and Staphylococcus. Bacteria belonging to class Bacilli, order Lactobacillales, include Enterococcus, Lactobacillus, Lactococcus, Leuconostoc,

Pediococcus and Streptococcus. In some embodiments, the Firmicutes belong to class Clostridia (e.g., order Clostridiales, Halanaerobiales or Thermoanaerobacteriales). Bacteria belonging to class Clostridia include, for example, Acetobacterium, Clostridium,

Eubacterium, Heliobacterium, Helio spirillum, Megasphaera, Pectinatu, Selenomonas, Zymophilus, Sporohalobacter and Sporomusa. In some embodiments, the Firmicutes belong to class Mollicutes (e.g., order My coplasmatales, Entomoplasmatales, Anaeroplasmatales or Acholeplasmatales). In some embodiments, a preterm infant may be identified as at risk for NEC (e.g., NEC-I), if the infant has a high relative abundance of Enterococcus and/or Staphylococcus.

In some embodiments, a preterm infant may be identified as at risk for NEC (e.g., NEC-II) if the relative abundance of the microbiota represents an elevated level of

Proteobacteria, as compared to the levels of Proteobacteria represented by a reference relative abundance of microbiota. The Proteobacteria are a phylum of bacteria, most of which have Gram-negative cell wall structure. In some embodiments, the Proteobacteria belong to class Alpha Proteobacteria (e.g., order Caulobacte rales, Kordiimonadales, Parvularculales, Rhizobiales, Rhodobacterales, Rhodospirillales, Rickettsiales or Sphingomonadales), Beta Proteobacteria (e.g., order Burkholderiales, Hydro genophilales, Methylophilales,

Neisseriales, Nitrosomonadales, Rhodocyclales or Pwcabacteriales), Gamma Proteobacteria (e.g., order Acidithiobacillales, Aeromonadales, Alteromonadales, Cardiobacteriales, Chromatiales, Enterobacteriales, Legionellales, Methylococcales, Oceanospirillales, Pasteurellales, Pseudomonadales, Thiotrichales, Vibrionales or Xanthomonadales), Delta Proteobacteria (e.g., order Proteobacteria, Bdellovibrionales, Desulfobacterales,

Desulfovibrionales, Desulfurellales, Desulfarcales, Desulfuromonadales, Myxococcales or Syntrophobacterales) or Epsilon Proteobacteria (e.g., order Campylobacterales or

Nautiliales). In some embodiments, a preterm infant may be identified as at risk for NEC (e.g., NEC-II) if the infant has a high relative abundance of Enterobacter and/or Escherichia.

In some embodiments, a preterm infant may be identified as at risk for NEC if the microbiota of the infant lacks Propionibacteria (belonging to the phylum Actinobacteria). Surprisingly, it was discovered that NEC cases may be characterized by a lack of Propionibacteria in the microbiota of the preterm infants, including those cases preceded by Firmicutes dysbiosis and those cases preceded by Proteobacteria dysbiosis.

The relative abundance of microbiota may be obtained from a stool sample. The stool sample may be collected prior to diagnosis of NEC. The sample may be obtained from the preterm infant during postnatal day 10 to postnatal day 16. For example, the stool sample may be obtained on postnatal day 10, postnatal day 11, postnatal day 12, postnatal day 13, postnatal day 14, postnatal day 15 or postnatal day 16. More than one sample may be collected and analyzed. The samples may be collected on the same day or on different days.

To practice any of the methods described herein, the level(s) of one or more of the metabolite biomarkers or microbial biomarkers noted above in a tissue sample of a candidate subject (e.g., a urine sample or a stool sample) is determined by performing a method known in the art (e.g., those described herein) or based on medical records of that candidate subject. When necessary, the data thus obtained can be normalized against the level of an internal control. The normalized level(s) of the metabolite biomarker(s) or the microbial marker(s) can then be compared to the level(s) of the same biomarker(s) of a reference level, which can be obtained from a control tissue sample to determine whether the subject has or is at risk for developing NEC or the subtype of NEC.

The reference level can be determined based on the level of the same metabolite and/or microbial biomarker from a matched subject who is free of NEC. Alternatively, the reference level can be obtained from a pool of NEC-free subjects. Optionally, these NEC- free subjects match with the test subject in, e.g., age, gender, and/or ethnic background.

Preferably, the control tissue sample and the tissue sample examined in the methods described here are of the same type.

When necessary (e.g., when more than one metabolite biomarkers and/or microbial biomarkers are investigated), the levels of these biomarkers can be processed by, e.g., a computational program to generate a profile (e.g., a metabolite signature, a microbial signature, or a combined metabolite and microbial signature), which can be represented by a number or numbers, that characterize the level of the biomarkers. The levels of the same biomarkers from the control subject(s) can be processed by the same method to generate a reference profile representing the level of these biomarkers of the control (a reference level). The profile of the test subject can be compared with the reference profile to determine whether the test subject has or is at risk for NEC (or a subtype of NEC) development. Various computational programs can be applied in the methods of this disclosure to aid in analysis of expression data. Examples include, but are not limited to, Prediction Analysis of Microarray (PAM; see Tibshirani et al., PNAS 99(10):6567-6572, 2002);

Plausible Neural Network (PNN; see, e.g., US Patent 7,287,014), PNNSulotion software and others provided by PNN Technologies Inc., Woodbridge, VA, USA, and Significance Analysis of Microarray (SAM).

Treatment of NEC

When a subject is diagnosed by any of the methods described herein as having or at risk for developing NEC, this subject could be subjected to a suitable treatment for NEC, e.g., those known in the art. Non-limiting examples of treatments for NEC include antibiotics, cessation of enteral feedings, fluid resuscitation, total parenteral nutrition, and/or surgery (e.g. , resection of the necrotized tissue and optionally colostomy, which may be a reversible colostomy).

The term "treating" as used herein refers to the application or administration of a composition including one or more active agents to a subject, who has NEC, a symptom of NEC, or a predisposition toward NEC, with the purpose to cure, heal, alleviate, relieve, alter, remedy, ameliorate, improve, or affect the disease, the symptoms of the disease, or the predisposition toward the disease. An "effective amount" is that amount of an anti-NEC agent that alone, or together with further doses, produces the desired response, e.g. eliminate or alleviate symptoms, prevent or reduce the risk of flare-ups (maintain long-term remission), and/or restore quality of life. The desired response is to inhibit the progression of the disease. This may involve only slowing the progression of the disease temporarily, although more preferably, it involves halting the progression of the disease permanently. This can be monitored by routine methods or can be monitored according to diagnostic and prognostic methods discussed herein. The desired response to treatment of the disease or condition also can be delaying the onset or even preventing the onset of the disease or condition.

Such amounts will depend, of course, on the particular condition being treated, the severity of the condition, the individual patient parameters including age, physical condition, size, gender and weight, the duration of the treatment, the nature of concurrent therapy (if any), the specific route of administration and like factors within the knowledge and expertise of the health practitioner. These factors are well known to those of ordinary skill in the art and can be addressed with no more than routine experimentation. It is generally preferred that a maximum dose of the individual components or combinations thereof be used, that is, the highest safe dose according to sound medical judgment. It will be understood by those of ordinary skill in the art, however, that a patient may insist upon a lower dose or tolerable dose for medical reasons, psychological reasons or for virtually any other reasons. Kits for use in NEC diagnosis and/or prognosis

Also within the scope of this disclosure are kits for use in diagnosing NEC in a subject (e.g., a preterm human infant). Such a kit can comprise reagents for determining the level(s) of one or more of the biomarkers described herein for use in the diagnostic methods also described herein. For example, the kit can contain antibodies specific to one or more of the metabolite biomarkers or more or more oligonucleotides that are specific to the 16s rRNA of a specific microbial marker. Any of the kits described herein can further comprise an instruction manual providing guidance for using the kit to perform the diagnostic/prognostic methods.

Without further elaboration, it is believed that one skilled in the art can, based on the above description, utilize the present invention to its fullest extent. The following specific embodiments are, therefore, to be construed as merely illustrative, and not limitative of the remainder of the disclosure in any way whatsoever. All publications cited herein are incorporated by reference for the purposes or subject matter referenced herein. Example 1: Biomarkers for NEC, Including Type I and Type II NEC

In this study, banked stool and urine samples collected prior to disease onset from infants <29 weeks gestational age were analyzed, including 11 infants who developed NEC (cases) and 29 matched controls who survived free of NEC. Stool bacterial communities were profiled by 16S rRNA gene sequencing, and urinary metabolomic profiles were assessed by NMR.

Results obtained from this study show that, during postnatal days 4 to 9, samples from infants who later developed NEC tended towards lower alpha-diversity (Chaol index, p=0.086) and lacked Propionibacterium (p=0.009) compared to controls. Furthermore, NEC was preceded by distinct forms of dysbiosis. During days 4 to 9, samples from four NEC cases were dominated by members of the Firmicutes (median relative abundance: >99 vs <17% in the remaining NEC and controls, p<0.001). During postnatal days 10 to 16, samples from the remaining NEC cases were dominated by Proteobacteria, specifically Enterobacteriaceae (median relative abundance >99% vs 38% in the other NEC cases and 84% in controls, p=0.01). NEC preceded by Firmicutes dysbiosis occurred earlier (onset, days 7 to 21) than NEC preceded by Proteobacteria dysbiosis (onset, days 19 to 39). All NEC cases lacked Propionibacterium and were preceded by either Firmicutes (>98% relative abundance, days 4 to 9) or Proteobacteria (>90% relative abundance, days 10 to 16) dysbiosis, while only 25% of controls had this phenotype (predictive value 88%, p=0.001). Analysis of days 4 to 9 urine samples found no metabolites associated with all NEC cases, but alanine was positively associated with NEC cases that were preceded by Firmicutes dysbiosis (p<0.001) and histidine was inversely associated with NEC cases preceded by Proteobacteria dysbiosis (p=0.013). A high urinary alanine:histidine ratio was associated with microbial characteristics (p<0.001) and provided good prediction of overall NEC (predictive value 78%, p=0.007).

Materials and Methods

Preterm Infant cohort. All infants in this study were part of an ongoing larger study of the preterm microbiome (also referred to as microbiota) and determinants of NEC, which was conducted in two level III neonatal intensive care units (NICUs). Enrollment criteria included being free of congenital anomalies and survival free of NEC in the first week of postnatal life. After Institutional Review Board approval and parental informed consent, standardized clinical data were collected following the NICHD Neonatal Research Network protocol until discharge from hospital. All cases were reviewed by a senior neonatologist.

A total of 35 preterm infants <29 weeks gestational age and <1200 g birth weight were included in this study. The primary analysis included 11 infants who developed NEC and 21 control infants. As a secondary comparator, three non-NEC deaths attributed to respiratory distress syndrome or suspected infection were included. Non-NEC deaths were included as they represent a competing outcome with NEC during hospitalization in the

Neonatal Intensive Care Unit (NICU), and clinical studies of probiotics and antibiotics report similar associations with NEC and death⁷' ⁸' ¹⁴ suggesting the possibility that they may be similarly related to the intestinal microbiome. Controls were consecutively enrolled infants from the same NICU who survived free of NEC or sepsis. NEC was defined as modified Bell's stage II or III.³⁰

The infant microbiome was analyzed by postnatal time periods, days 4 to 9 and 10 to 16. These periods were selected because the extremely premature infants in this study did not stool consistently and did not stool prior to day of life 4. Each period included two planned sample collections, increasing the chances of having sample available from each infant during the interval. Stool was collected from soiled diapers and urine samples were collected using cotton balls in the infant diaper. Upon collection, samples were immediately refrigerated in the NICU and transported to the study laboratory daily. Stool was scraped into a tube containing thioglycollate, and frozen at -80° C until DNA extraction was performed. Upon removal from the cotton ball, urine samples were frozen at -80° C without additive until metabolomic analysis was performed. All sample collection and storage utilized preprinted, bar-coded labels.

In the few instances when two stool samples were available per infant in a collection period, statistical independence was assured by including only the earlier sample. Further, only samples collected prior to diagnosis of NEC were included. These criteria resulted in inclusion of 58 stool samples for analysis, 18 from the 11 NEC cases, 37 from the 21 control infants, and 3 from the 3 non-NEC deaths. Urine samples were collected during days 4 to 9, and all available samples collected prior to NEC onset in that time period were analyzed by

NMR. 31 A total of 60 urine samples, 18 from 11 NEC cases, 36 from 20 controls, and 6 from non-NEC deaths, were included in initial analyses. Final analysis of urinary biomarkers was restricted to the first urine sample collected in the day 4 to 9 period.

DNA Extraction. Bacterial DNA was extracted from infant stool samples using one of two methods: Phenol-chloroform or the QiaAmp DNA stool kit (Qiagen Sciences,

Germantown, MD). For the phenol-chloroform method, samples were thawed, centrifuged and supernatant removed. 0.2 grams of stool was transferred to a 2 mL screw-cap tube containing 0.3 g of 1mm zirconia beads. 1 mL TE (Tris, ethylenediaminetetraacetic acid (EDTA)) buffer was added and the sample resuspended. 150 μΐ_^ of buffer saturated phenol was added and the sample bead beat for 3 minutes at 4°C. Sample was then allowed to sit 1 minute at 4°C before extraction with 150 chloroformdsoamyl alcohol (24: 1). Samples were centrifuged for 5 minutes at 15,700 g at 4°C and the resulting aqueous layer was transferred to a clean tube and again extracted with an additional 150 phenol and 150 Chloroformdsoamyl alcohol. Samples were then centrifuged for 2 minutes at 15,700 g at 4°C. The aqueous layer was subjected to two additional chloroformdsoamyl alcohol extractions. The resulting aqueous layer was transferred to a clean microfuge tube a final time and 4 of 5 mg/mL glycogen was added and mixed. This was followed by the addition of 1/10 volume of 3 M sodium acetate and 2 volumes of 70% cold (-20C) ethanol. The samples were mixed and then centrifuged at 15,700 g for 5 minutes at 4°C. The supernatant was discarded and the pellet resuspended in 100 TE buffer. For the extractions using the QiaAmp DNA stool kit, samples were thawed and 0.2 grams of stool were transferred to a bead beating tube containing 0.3 grams of 0.1 mm glass beads, 1.4 mL of buffer ASL (Qiagen buffer) was added, and bead beating was conducted for 3 minutes on the homogenize setting. The suspension was then heated at 70°C for 5 minutes, and the manufacturer instructions followed.

16S rDNA analysis. The V3 to V5 window of the 16S rRNA gene was amplified and sequenced using 454 FLX Titanium sequencing. 32 A total of 1.3 M resulting sequences were processed using a data curation pipeline implemented in mothur, 33 complemented by

UCHIME for chimera detection.³⁴ This processing is detailed previously, ³⁵ with the modification that Abundant OTU was replaced by Newbler for assembly-based error reduction³⁶ and followed by mothur's implementation for operational taxonomic unit (OTU) clustering (parameters: method=average, cutoff=0.03, precision=1000). The mean read count per sample was 4,989. Representative sequences per OTU were classified with the MSU

RDP classifier v2.2 37 using the Greengenes taxonomy. 38 The 16S rRNA gene sequences obtained from the study samples were assigned to a total of 411 distinct OTUs. OTUs were then assigned phylogenetic classifications, typically to the genus level.

Metabolomic analysis. Urine samples were thawed on ice immediately prior to preparation for NMR analysis. A 1 mL aliquot of each sample was centrifuged for 10 minutes at 2655xg, 350 mL of clarified urine pipetted into a 1.5 mL microcentrifuge tube, and 350 mL of buffer added to each sample. 600 mL of the urine/buffer mixture was then pipetted into a 5 mm NMR tube. All NMR experiments were carried out on a Bruker Avance™ III spectrometer operating at 600 MHz 1H frequency equipped with a room temperature 5 mm triple resonance probe with inverse detection and controlled by TopSpin 3.0 (Bruker Biospin, Germany). Experiments were conducted at 298 K. Data were collected using a spectral width of 20.0 ppm. Three 1H NMR experiments, optimized by Bruker (Bruker BioSpin, Germany) for use in metabolomic studies, were run on all samples: a standard one-dimensional (ID) presaturation (zgpr), a ID first increment of a NOESY (noesygpprld), and a CPMG (cpmgprld), however, only the CPMG, which produced superior baselines for analysis, was used for metabolomic analysis. The transmitter offset frequency (01) was adjusted to obtain optimal water suppression. The 90° pulse widths, determined for every sample using the automatic pulse calculation feature in TopSpin, were between 10 and 12 μ8. Water suppression in all experiments was achieved by irradiation of the water peak during the recycle delay. ID zgpr 1H NMR spectra were collected to assess the shim quality which was considered acceptable when the line width was <lHz and the line shape enabled detection of resolved C13 satellites for the TSP internal standard. The CPMG experiment used 64 transients with 4 dummy scans, 46280 points per spectrum giving an acquisition time of 1.87 s, a T₂ filter loop of 128 with an echo time of 1 ms, apodized using - 0.01 Hz of exponential line broadening, and a 4 s recycle delay. All NMR spectra were phased, baseline corrected, and subjected to chemical shift registration relative to TSP in TopSpin 3.0 (Bruker BioSpin, Germany).

Statistical methods. Differences in clinical characteristics of cases and controls were tested using the Fisher's Exact test for categorical variables and analysis of variance, t- tests or Kruskal-Wallis for continuous variables, as appropriate. The Kruskal-Wallis test was used to compare the relative abundance of distinct taxonomic units. Data were analyzed using R, ³⁹ LEfSe, ⁴⁰ QIIME,⁴¹' ⁴² and SAS⁴³ Linear discriminant analysis (LDA) effect size (LEfSe) was used to identify the phylogenetic features that differed significantly between all NEC cases and controls and later between NEC sub-types and controls; this program uses Kruskal- Wallis tests to identify differences in abundance between groups at the alpha level of 0.05 ⁴⁰ Results

(A) Microbial community analysis

Alpha-diversity was first examined, ^M' ⁴⁵ which was analyzed using the Simpson diversity index and Chaol richness index. Samples collected prior to NEC onset were compared to controls without elimination of rare reads but after rarefying (standardizing) to 2000 sequence reads per sample. To examine the beta diversity of microbial communities, non-metric dimensional scaling (NMDS) was used, ⁴⁶'⁴⁷ undertaken in R, to ordinate the microbial communities of samples based on weighted UniFrac distance calculated by QIIME. To improve signal to noise and reduce random error, rare OTUs were excluded from the beta diversity analyses. Rare OTUs were defined as those detected in only one sample, or with less than five sequences in the overall dataset.¹⁰ This resulted in 99 distinct OTUs that were included for analysis. Ordination using NMDS was undertaken with several random starts to avoid entrapment in local optima. The results were centered and scaled, so that the variance was maximized along the first NMDS axis. This method of ordination was chosen because NMDS measures the closeness of fit (stress) based on ranking of the dissimilarity of values, with no assumption of multivariate normality, and is a powerful and flexible method that handles sparse, non-parametric data well. Use of the weighted UniFrac distance metric was selected, as it accounts for the relative abundance and relatedness of taxa, rather than merely their presence or absence. Selection criteria for the number of dimensions in NMDS analysis were based on the protocol used in PC-ORD 6 software.⁴⁷ For each ordination, a cluster analysis was performed on the UniFrac distance matrix using the Ward minimum variance method 48 to objectively identify clusters of samples with similar microbial composition. In this method, the distance between two clusters is the analysis of variance sum of squares between the clusters summed over all the variables. At each step, the within-cluster sum of squares is minimized over all partitions obtainable by merging clusters from the previous generation in order to maximize the likelihood at each level of the hierarchy. The number of clusters is determined by examining the scree plot resulting from the pseudo F and T plots. SAS PROC Cluster procedure was used to conduct the clustering analysis.

Metabolomics allows identification of distinct patterns of small molecules generated during both host and microbial cellular metabolism. 25-^"29 Urinary metabolomics was thus undertaken in search of surrogate biomarkers of dysbiosis and additional clues regarding the microbially-distinct NEC cases in relation to controls and to each other. Urinary metabolite data was analyzed using Principal Components Analysis (PCA) as implemented in ΑΜΓΧ (Bruker Biospin, Billerica, MA). NMR spectra were prepared for PCA using manual binning to avoid splitting of peaks. Individual metabolites were compared between NEC cases and controls using a t-test as implemented by Proc MULTTEST in the SAS 9.2 software (SAS Institute, Cary, NC). The Benjamini and Hochberg procedure⁴⁹ was used to control for multiple comparisons, using an adjusted p-value of <0.05 ⁴⁹ After identification of significant metabolites, generalized estimating equations (GEE) were applied to adjust for multiple samples per subject and control for potential confounding factors. Only metabolites that were significant at <0.05 in both analyses were considered significant. A total of eight samples analyzed by NMR were excluded based on poor sample quality or as outliers based on the Hotelling T2 method.⁵⁰ Preterm infant cohort. The characteristics of NEC cases versus controls are shown in Table 2. Controls were generally well matched to NEC cases on clinical factors, and did not differ in regard to birth weight (median 850 grams), gestational age (median 26 weeks), race (37% black), gender (51% female), mode of delivery (66% Cesarean section), maternal antibiotic use at the time of delivery (51%), or infant antibiotic use > 5 days in the first week (34%). All study infants were fed mother's own milk or human donor milk; the timing and degree of feeding was not significant in relation to NEC. Only primiparity differed between NEC cases (64%) and

Table 2. Characteristics of study infants

*NEC vs controls, primiparity p=0.05 controls (24%, p=0.05), but was not significantly associated with microbial composition. Of the 11 NEC cases, eight were Bell's stage II and three were Bell's stage III (surgically treated). Four of the NEC cases died, and three of the NEC cases developed sepsis (two Klebsiella isolates, one coagulase negative Staphylococcus isolate). The three non-NEC deaths included for comparison were white, male, younger (24 and 25 weeks gestational age) and smaller (<850 grams birth weight) than the other infants, but otherwise unremarkable (data not shown).

Dominant organisms. Consistent with previous studies in preterm infants, 12 ' 51-^"53 the dominant phyla were Proteobacteria and Firmicutes, with a minor contribution (1% to 2%) from Bacteroidetes and Actinobacteria. The most common genera were, in order of relative abundance: Enterobacter, Staphylococcus, Escherichia, Enterococcus, Leuconostoc, Lactococcus, Streptococcus, and Clostridia. The first four genera accounted for more than 90% of the microbial sequence reads. Pseudomonas, another Proteobacteria often associated with NEC or sepsis in preterm infants, occurred in 31% of samples, while together,

Lactobacillaceae and Bifidobacteriaceae, beneficial Gram-positive organisms (respectively, of the phyla Firmicutes and Actinobacteria), were present in only 19% of samples.

At the level of phyla, controls had a median relative abundance of approximately 80% Proteobacteria (Gram-negative organisms) and 20% Firmicutes (Gram-positive organisms), with a small proportion of Bacteroidetes and Actinobacteria; this pattern was remarkably stable over the first few weeks of life. Furthermore, most of the sequences contributing to the relative abundance of these large phyla came from only a few host- associated genera, Enterococcus and Staphylococcus for Firmicutes and Enterobacter and Escherichia for Proteobacteria. In infants who later developed NEC, microbial community composition differed sharply with median relative abundance of Proteobacteria less than 40% and

Firmicutes -60% during postnatal days 4 to 9. A dramatic difference was then observed in samples from NEC cases such that Proteobacteria comprised above 90%, while Firmicutes comprised less than 10% of the microbial communities (FIG. 1). In the three non-NEC deaths (data not shown), Firmicutes strongly dominated, comprising 70 to 90% of microbial communities, with most of the remaining community composition comprised of

Proteobacteria.

Systematic comparison of all NEC samples and all control samples collected between postnatal days 4 to 16 found no significant differences in microbial composition. NEC and control samples were then compared by week, and the only significant difference occurred in the relative abundance of Propionibacterium, a genus of the phylum Actinobacteria.

Propionibacterium includes both skin and intestinal tract colonizing organisms; members have demonstrated probiotic as well as pathogenic potential.⁵⁴' ⁵⁵ During days 4 to 9, Propionibacterium was identified in samples from 10 (56%) of the 18 control infants, but none of the 9 infants who later developed NEC (p=0.01, FIG. 2, Panel A). No difference was found in the relative abundance of Propionibacterium between NEC and control samples collected from days of life 10 to 16.

Alpha diversity. During postnatal days 4 to 9, infants who later developed NEC tended to have samples with lower alpha diversity than controls as measured by Chaol index (median, 9.2 for NEC, 18.4 for control samples; Kruskal-Wallis, p=0.086) with a similar pattern for Simpson index (p=0.221) (FIG. 2, Panel B). After day of life 9, NEC samples continued to trend towards lower median alpha-diversity than controls, but no significant differences were found by either index.

Dysbiosis-identified sub-types of NEC. For all study infants, the relative similarity of microbial communities between samples (beta-diversity) was then examined by calculating their weighted UniFrac distances, and visualized using non-metric multi-dimensional scaling ordination (NMDS). Based on ordination of day 4 to 9 samples (FIG. 3, Panel A), application of the Ward minimum variance method identified four clusters, designated as I though IV, indicating microbial community similarity. Of these, only Clusters I and II included NEC cases. Cluster-I consisted of samples from 4 NEC cases (hereafter noted as NEC-I), 2 non-NEC deaths, and 2 controls (FIG. 3, Panel B), and were characterized by dominance of organisms of the phylum Firmicutes, class Bacilli. Cluster- II samples consisted of the remaining 5 NEC cases (hereafter noted as NEC-II) as well as 12 of the 18 controls, and were characterized by dominance of organisms of the phylum Proteobacteria, family Enterobacteriaceae.

Within Cluster-I, the genera Enterococcus and Staphylococcus, taxa representing different orders of Bacilli, comprised 98% or more of the microbial community of samples from NEC-I infants. In the two non-NEC deaths found in Cluster-I, the same taxa constituted 80% and 95% of their microbial communities, and in the two control samples, 62% and 73%. Comparing the NEC cases between clusters, the relative abundance of Firmicutes

(specifically, Bacilli) was significantly higher in samples from NEC-I (type I) than NEC-II (type II) cases (median: 99.3%, NEC-I vs 17%, NEC-II, p=0.014). Further, NEC-I had significantly greater relative abundance of Firmicutes, specifically, Bacilli (p=0.001), when compared to all controls. Conversely, the relative abundance of Proteobacteria, family Enterobacteriaceae, was significantly higher in samples from NEC-II than NEC-I cases (median: 83%, NEC-II vs 0.4%, NEC-I, p = 0.014), but NEC-II samples did not differ significantly in microbial composition compared to all controls during days 4 to 9.

Ordination of samples from days 10 to 16 (FIG. 4, Panel A) and application of the Ward minimum variance method identified three microbial community clusters (A, B, and D) and an outlier (C). The NEC-I infants that had tightly clustered in samples from the first 4 to

9 days were dispersed across Clusters A, B and D in this ordination, with no discernible similarity. In contrast, consistent with the days 4 to 9 ordination, all of the NEC-II cases were found in a single cluster (Cluster- A) during days 10 to 16. Cluster- A also included one of the dispersed NEC-I cases, the composition of which was 83% Enterobacter. Cluster-A thus included 7 (78%) of the 9 NEC cases and 12 (63%) of 19 controls, and was characterized by preponderance of Proteobacteria, specifically, the family Enterobacteriaceae (FIG. 4, Panel B). Compared to all control samples during days 10 to 16, the six NEC-II/Cluster-A cases had a significantly elevated relative abundance of Proteobacteria (specifically,

Enterobacteriaceae, p=0.010). All six NEC-II/Cluster-A infants vs. 8 (42%) of 19 control samples were comprised 90% or more of Proteobacteria (p=0.020). The composition of samples was systematically compared, and found, similar to the prior week, that during days

10 to 16, NEC-I had significantly higher relative abundance of Firmicutes than NEC-II, specifically, Bacilli (median: 33%, NEC I vs 0.3%, NEC II, p=0.020). However, the relative abundance of these taxa in NEC-I cases during days 10 to 16 (median: 33% Firmicutes or Bacilli) were not as extreme as that observed earlier (median: >98% Firmicutes or Bacilli during days 4 to 9). In contrast, NEC-II cases became even more dominated by

Proteobacteria, specifically, the Enterobacteriaceae (median: 99.6%, NEC-II vs 38%, NEC I, p=0.020). Furthermore, during days 10 to 16, NEC-II samples also had significantly greater relative abundance of Proteobacteria when compared to all controls (median: 99.6%, NEC-I vs 84%, controls, p=0.01).

In summary, four NEC cases were classified as NEC-I, all of whom were found as part of Cluster-I, uniquely characterized by >98% relative abundance of Firmicutes, class Bacilli, during postnatal days 4 to 9. Six NEC cases were classified as NEC-II, all of whom were found as part of Cluster-A during days 10 to 16 as well as Cluster- II during days 4 to 9. These NEC-II cases all were composed of >90% Proteobacteria, family Enterobacteriaceae, in samples from postnatal days 10 to 16. One NEC infant (subject 16) lacked sample from days 10 to 16, and could not be formally classified, but followed the pattern of NEC-II based on their days 4 to 9 sample, which was dominated by Escherichia and found within Cluster- II. All three non-NEC deaths were characterized by early Firmicutes dysbiosis, similar to that of NEC-I. One of these non-NEC deaths (subject 40) lacked sample from days 4 to 9, but was considered to be a high Firmicutes dysbiosis based on their days 10 to 16 sample, which was predominantly composed of Staphylococcus. The primary ordinations of beta- diversity (FIGs. 3 and 4) included all samples and sequence reads, but ordination using rarefied samples (FIG. 6) found the same pattern as that shown in FIGs. 3 and 4.

The alpha-diversity of microbial communities was then re-analyzed in relation to these NEC sub-types, as before using the full set of OTUs without eliminating rare sequence reads but rarefying samples to 2000 sequence reads per sample. No significant differences were found in NEC sub-types by either Simpson or Chaol. The clinical characteristics of the NEC-I and NEC-II infants were then compared. NEC-I occurred between postnatal days 7 and 21, while NEC-II occurred between postnatal days 19 and 39 (p=0.086, Kruskal-Wallis test). It is noteworthy that the non-NEC deaths, which clustered with NEC-I samples in the ordination of samples from days 4 to 9, had a similarly high Firmicutes dysbiosis in the first week of life, and that these deaths occurred between days 9 to 17, during the same postnatal period as the NEC-I cases. The two NEC sub-types were then compared in relation to each variable listed in Table 2. See also Table 3 below.

Table 3. Characteristics of NEC case types

+ Comparison between NEC types significant for maternal antibiotic use, p=0.015, Fisher's exact * Difference between NEC types exhibits a statistical trend for primiparity and PDA, p=0.089 One statistical difference in the two NEC sub-types was in the administration of antibiotics to the mother at the time of delivery. No NEC-I, but 6 (86%) of the 7 NEC-II infants, were born of mothers who were given antibiotics at the time of delivery (p=0.015). While different delivery modes did not readily explain the association between NEC-II and maternal antibiotic use, the beta-diversity of microbial communities was examined in relation to delivery mode and NEC or control status, and observed a tendency for NEC-I to cluster with C-section delivery, and NEC-II to cluster with vaginal delivery (FIG. 7). For two other clinical factors, patent ductus arteriosis (PDA, a heart problem of preterm infants that has been linked to later-onset NEC) and primiparity, a trend (p=0.089) towards a different distribution in NEC-I and NEC-II cases was observed. Each factor was independently found in only 1 (25%) of 4 NEC-I infants versus 6 (86%) of 7 NEC-II infants. No other clinical characteristics appeared to differ between the NEC sub-types, nor in comparison to controls.

(B) Urinary metabolomic analysis

Principal components analysis did not demonstrate any qualitative clustering in the set of urinary metabolites in relation to all NEC cases, NEC-I or NEC-II versus controls, nor the NEC sub-types in relation to each other. Individual urinary metabolite values, which were parametric in distribution, were compared using t- tests corrected for multiple comparisons. No urinary metabolites differed significantly between all NEC cases and controls. However, three metabolites, alanine, pyridoxine or 4-pyridoxate, and histidine, significantly

distinguished NEC-I and NEC-II from each other as well as one of the NEC sub-types from controls (FIG. 5 and Table 4 below).

Table 4. Urinary Metabolite Levels in NEC Sub-Types

Urinary metabolites, including alanine, pyridoxine, histidine, and tyrosine, were identified in samples collected DOL 4 to 9. Differences were observed between case types and between case type and controls, compared using generalized estimating equation (GEE) models. Data are presented as β-coefficients, 95% confidence intervals (CI) and p-values.

Alanine was significantly (p<0.001) higher in NEC-I versus NEC-II and NEC-I versus control samples though the metabolite did not differ between all NEC versus control samples. Pyrixodine followed a pattern similar to alanine, though not as significant. The remaining metabolite, histidine, differed in pattern and was significantly lower in NEC-II samples than controls (p=0.01) or NEC-I samples (p<0.001) (FIG. 5).

Alanine, pyridoxine, and histidine are commonly synthesized by bacterial enzymes, as documented by KEGG⁵⁶ and may be considered plausible biomarkers of bacterial dysbiosis. The relationship of these urinary metabolites to microbial community characteristics in the dataset was then analyzed, using only the first urine sample collected between days 4 to 9 from 28 infants (9 NEC cases, 2 non-NEC deaths, and 17 controls) who also had a stool sample analyzed from days of life 4 to 9. Alanine (Table 5) was significantly associated with characteristics of the intestinal microbial community. Alanine levels were most strongly associated (p<0.001) with Cluster-I samples identified in the days 4 to 9 ordination, including the NEC-I cases, non-NEC deaths and the controls in that cluster. Alanine was also directly correlated with the relative abundance of Firmicutes (p=0.009), and inversely correlated with the relative abundance of both Proteobacteria (p=0.027) and Propionibacterium (p=0.015). Alanine was not associated with the number of days elapsed between collection of the urine sample and case onset, and thus appeared to be associated with dysbiosis rather than host response. Pyridoxine was not as strongly associated with microbial community

characteristics, and its association was explained by correlation with alanine (data not shown). Histidine was not independently associated with microbial community

characteristics (Table 5), but among the 11 NEC cases had a strong inverse association with the number of days between collection of the urine sample and case onset (Spearman's r=- 0.68, p=0.020). Thus, histidine appeared to be associated with host response rather than dysbiosis. Because alanine and histidine were observed to have distinct and possibly complementary associations in relation to NEC-I and NEC-II (Table 5 and FIG. 5), ratio of alanine to histidine in relation to NEC outcomes and dysbiosis was examined. Remarkably, the ratio was positively associated with overall NEC (Kruskal-Wallis, p=0.001), was inversely associated with the relative abundance of Propionibacterium (Spearman's r=-0.57, p=0.002), and did not differ by NEC sub-type nor in relation to time between sample collection and disease onset.

Table 5. Association between urinary metabolite values, microbial community characteristics and NEC in 28 infants*

* Analyses include 9 NEC, 2 deaths and 17 controls, except analysis of NEC, NEC-I, and NEC-II, for which the non-NEC deaths were removed as a competing cause. Metabolites measured as normalized peak intensity. Significant values are bolded.

* Urine and stool samples collected between postnatal days 4 to 9

Principal components analysis did not find any distinctive clustering in the set of urinary metabolites in relation to all cases and controls, case typel or 2 and controls, or the case types in relation to each other. Urinary metabolites were then compared using t-tests corrected for multiple comparisons. While only one urinary metabolite, alanine, modestly differed (p=0.05) between all cases and controls, four metabolites, alanine (p<0.001), pyridoxine or 4-pyridoxate (p<0.001), histidine (p=0.002) and tyrosine (p=0.007), significantly distinguished case types 1 and 2 from each other; indeed, all four metabolites were significantly higher in the days 4 to 9 urine of infants who later became a type 1 compared to type 2 case (figure 4, panel C and Supplementary Information, table 4): Further, two of the metabolites (alanine and pyridoxine or 4-pyridoxate) were significantly (p<0.001) higher in case type 1 than all other infants, while three of the metabolites (alanine, histidine and tyrosine) were significantly lower in case type 2 compared with all other infants. The correlations of these four metabolites with each other ranged from 0.4 to 0.6.

Predictive biomarkers. To systematically evaluate the signatures identified above as potential predictors of NEC onset, area under the receiver operating curve (ROC) analysis was conducted, beginning with defined microbial colonization characteristics (Table 6).

Table 6. Microbial and metabolite predictors of NEC

p-values based on Fisher's exact test. The variations in denominator are due varying sample availability.

First, high Firmicutes dysbiosis (>98% relative abundance in samples from days 4 to 9) was examined, which occurred in 4 of 9 NEC cases versus 0 of 18 controls (p=0.007) who had sample during days 4 to 9. This measure had a good predictive value (72%) and optimal specificity (100%). Next, the absence of Propionibacterium in samples from days 4 to 9 as a potential predictor was examined, which occurred with all NEC cases but only 44% of 18 controls (p=0.009). While the predictive value of the absence of these Propionibacterium was good (78%) for prediction of NEC, with optimal sensitivity (100%), the specificity was poor (56%). High Proteobacteria dysbiosis in days 10 to 16 samples as a predictor of subsequent NEC was then examined. ROC analysis identified two cut-points for high Proteobacteria dysbiosis, > 90% or >98% relative abundance, both of which maximized the predictive value (62%), but neither was significant. The >90% cut-point was selected, as it included all NEC-II cases identified in Cluster- II/Cluster-A by ordination. The likelihood of developing NEC from having either form of dysbiosis was examined next. This analysis, limited to infants with samples in both time periods, found that all 7 NEC cases vs 8 (50%) of 16 controls (p=0.052) had a form of early dysbiosis, which increased predictive value to 75%. But, the highest predictive value (88%) was obtained from the combination of either form of dysbiosis and lack of Propionibacterium, which occurred in 7 of 7 NEC cases and 4 (25%) of 16 controls (p=0.001), and thus, had ideal sensitivity (100%) and good specificity (75%).

Urinary metabolites as surrogate markers for prediction of NEC were examined next. As noted above, alanine, pyridoxine and histidine alone were each significantly associated with either NEC-I or NEC-II but not overall NEC. However, when analyzed together, alanine and histidine were significantly associated with NEC (FIG. 5, Panel C and Table 5). ROC analysis found the optimal cut-point to be a urinary alanine to histidine ratio >4 (Table 6), which yielded a predictive value of 78%. Alanine to histidine ratio values above the cut- point occurred in 9 (82%) of 11 NEC cases and 5 (25%) of 20 controls (p=0.007), providing very good sensitivity (82%) and good specificity (75%).

DNA extraction. As DNA extraction methods can affect the results of 16S rDNA studies, a series of analyses were conducted to identify potential effects in the data.

Examination of extraction protocol in relation to beta-diversity found no influence on identification of community clusters (FIG. 8, Panel A). Examination of extraction protocol in relation to the relative abundance of bacteria at all taxonomic levels found taxa that differed by extraction method, but none of the taxa identified also differed significantly between NEC or NEC sub-types and controls (FIG. 8, Panel B). Finally, examination of extraction protocol in relation to alpha-diversity found no association with the Simpson index, but did find association with the Chaol index (p=0.01). Finally, the extraction protocol in regression models of alpha-diversity in relation to NEC was examined, and found that it did not influence the findings. These data support the conclusion that early differences occur in the microbiome of premature infants prior to onset of NEC, independent of extraction techniques.

This study revealed several factors that were significantly associated with subsequent risk of NEC, specifically, lack of Propionibacterium in the first week and two distinct forms of dysbiosis that occurred over the first two weeks of life: During days 4 to 9, the microbial community prior to onset of early NEC cases consisted predominantly (>98 ) of Firmicutes, specifically, class Bacilli, of which the dominant genera were Staphylococcus and

Enterococcus. This unique Gram-positive microbial signature was not shared by any of the control infants. The second clustered microbial phenotype occurred during 10 to 16 days of life preceding later onset NEC cases, and consisted of a Gram-negative Proteobacteria signature, specifically, family Entewbacteriaceae, of which the dominant genera were Enterobacter and Escherichia. The disparate timing and composition of these high Firmicutes and high Proteobacteria microbial signatures are intriguing, especially given their association with early versus late onset NEC or death.

The discovery of two forms of dysbiosis by metagenomic analysis was supported by metabolomic analysis, which identified differences among urine samples that were collected in the first week of life prior to case onset. These two '-omic' methods provide

complementary information. Urinary metabolomics is a sensitive method of identifying groups that differ in their intestinal bacterial colonization.²⁵'⁴⁴ Production and utilization of specific metabolites differs among colonizing bacteria, which in turn affects their

bioavailability to the host.²⁶ While metagenomic analysis of microbial DNA provides a comprehensive snapshot of bacterial composition, metabolomic comparison of microbial colonization phenotypes provides a snapshot of their differential metabolic activity. 27-^"29 In this study, at least three metabolites differed significantly between one of the dysbiosis types and controls. Alanine, higher in NEC-I as well as non-NEC deaths compared to control samples, is a non-essential amino acid that is ubiquitously incorporated into bacterial cell wall biosynthesis, a potential target of immune sensing.^58"60 As peptidoglycan constitutes most of the dry weight of Gram-positive organisms but only a small share of the dry weight of Gram-negative organisms, alanine seems a particularly promising candidate to

differentiate patients whose microbiome is strongly dominated by Gram-positive organisms.

Pyridoxine, also elevated in the urine of infants with early Gram-positive dysbiosis, is produced by bacteria in general⁶¹ and may reflect bacterial abundance or growth. The third metabolite that differed between the case types was histidine, a proteinogenic amino acid that was significantly lower in NEC-II infants compared with NEC-I infants and controls.

Urinary alanine was positively associated with the intestinal microbial community

composition and the relative abundance of Firmicutes, and negatively correlated with the relative abundance of Proteobacteria and Propionibacterium. Neither of the other two metabolites was independently associated with microbial community composition, but the ratio of alanine to histidine was significantly associated with NEC overall as well as with the relative abundance of Propionibacterium.

The preterm infants in this study generally lacked microbiota that are known to influence healthy immune development and oral tolerance, including Bifidobacterium, Bacteroides fragilis, and other commensal gut microflora.²³' ⁶²

Propionibacterium, a genus of the phylum Actinobacteria, was the only organism that differed significantly between all NEC cases and controls in this study. The organism was identified in the first postnatal week in about half of the controls but none of later NEC cases, suggesting a potential commensal role. The genus includes many species and strains that are used as probiotics by the dairy industry.⁵⁴ Other Propionibacterium commonly colonize the skin⁶³ and have been reported in breast milk.⁶⁴ These organisms are so named due to their production of propionic acid as well as other short chain fatty acids that have a beneficial role in intestinal health. The role of Propionibacterium in the intestinal colonization of infants is not known.

These findings provide important new insights, particularly regarding the role of Gram-positive intestinal dysbiosis in NEC that occurs by or before the third week of life. While older preterm infants tend to have earlier onset of NEC, neither infant gestational age nor birth weight explain these findings. All study infants were <29 weeks gestation and <1200 g birth weight, and neither variable differed between NEC and control infants nor between NEC cases with a microbial phenotype characterized as Firmicutes (Bacilli) or Proteobacteria (Enterobacteriaceae) dysbiosis.

The discovery of two specific forms of early microbial dysbiosis in the preterm infant that precede NEC or death provides a sharp focus for investigation of aberrant microbial- mucosal communication as part of the pathobiology leading to adverse outcomes. Further, the data indicate that characterization of early dysbiosis, along with the presence or absence of potentially probiotic organisms, may serve as non-invasive biomarkers that can together predict NEC in preterm infants. This study provides proof of concept approximately 80% prediction of NEC may be achieved by measures directly from fecal samples, or indirectly through surrogate biomarkers such as urinary metabolites over the first weeks of life. Prediction may be especially powerful using samples from the first week of life combined with samples from the second week of life.

Example 2: Biomarkers for NEC, including Typel, Type IIA, and Type IIB NEC

Unless otherwise stated, the methods used in this Example are the same as those described in Example 1.

In a follow-up study, 30 Cases of NEC, 60 controls, <29 weeks gestational age were tested. Stool and urine samples were collected at week 1, week 2, and week 3 post-birth for each subject. The relative similarity of microbial communities between samples (beta- diversity) was then examined by calculating their weighted UniFrac distances, and visualized using non-metric multi-dimensional scaling ordination (NMDS). Week 1, week 2, and week 3 post-birth samples were analyzed (Figure 17). Clusters were identified and designated NEC-I (high Firmicutes in first sample), NEC-IIA (sepsis prior to or concurrent with NEC; a subclass of type II NEC), and NEC-IIB (other; a subclass of type II NEC). NEC IIA and NEC IIB were determined to be sub-types of NEC II described in Example 1.

A. NEC IIA: Sepsis prior to or concurrent with NEC - high Enterobacteriaceae in

first sample

Of the 30 cases, 6 (20%) NEC were preceded by sepsis or sepsis occurred on the same day. One case of sepsis occurred after NEC, bit did not have the same early microbiota as the other cases. These were analyzed as a subgroup (NEC IIA), and a distinct pattern was evident. In the first stool sample, the 6 NEC cases preceded by or concurrent with sepsis had significantly higher relative abundance of phylum Proteobacteria (p=0.006, KW test) and family Enterobacteriaceae (p=0.004, KW test) compared to the 84 other infants (Figure 18). The median gestational age for the NEC IIA subgroup was 25 weeks and this sub-group was more likely to have maternal antibiotics given at the time of delivery.

B. NEC I: High Firmicutes in first sample

Of the 30 cases, 24 (removing NEC preceded by sepsis) and 60 controls were compared for levels of Firmicutes (comprised of Staph, Enterococcaceae, and Strep). Using an 85% cut-point, there were 11/24 NEC (46%) vs 6/30 (20% controls), p=0.001, Fisher's exact test. Thus, it was found that there was a significant increase in the phylum Firmicutes in the cases vs control. If all NEC cases were included in the analysis, there were 11/30 (37%) NEC, p=0.004. This subgroup was characterized by a trend to C-section and to earlier postnatal occurrence. This sub-group was also significantly older in gestational age: 27 weeks.

C. NEC IIB: "Other" - no evident difference in the first week sample

Some cases had no evident week 1 signature and were deemed the NEC IIB subcategory. The median in this sub-group was 25 weeks gestational age.

D. Microbial succession

Controls vs. NEC types were then examined for microbial succession, particularly in relation to the phylum Proteobacteria and family Enterobacteriaceae. It was found that phylum Proteobacteria and family Enterobacteriaceae differed from controls in each type of NEC (Figure 20). Between weeks 1 and 3, NEC types I and IIB had significantly different relative abundance of Enterobacteriaceae at baseline. However, in both circumstances, they increased in the relative abundance of Enterobacteriaceae between weeks 1 and 3 (Figure 20). The starting high Proteobacteria and family Enterobacteriaceae in sepsis-NEC and relative increase in Proteobacteria! Enterobacteriaceae in NEC types I and IIB are consistent with the widely held hypothesis that preterm infants have an excessive hyperinflammatory response to LPS-bearing organisms (Table 7, p=0.006. Excluding Sepsis-NEC, p=0.002).

Table 7. Postnatal weeks increase in Proteobacteria in relation to NEC

It was found that 87% Enterobacteriaceae in the first sample was not predictive of NEC as a whole, but was highly predictive of NEC preceded by sepsis {e.g., is predictive of sepsis, a risk factor for NEC, Table 8 and Table 9). Table 8. Association between high abundence of Enterobactenaceae and Type IIANEC

Table 9. Association between high abundence of Enterobacteriaceae and Type I NEC

High Firmicutes in the first sample was 68% predictive of NEC not preceded by sepsis (Tables 10 and 11, Figure 21).

Table 10. Association between high abundence of Firmicutes and NEC not preceded by sepsis

Table 11. Association between high abundence of Firmicutes and NEC not preceded by sepsis

E. Urinary Metabolites

As described in Example 1, the ratio of urinary alanine and histidine was identified as predictive of NEC. These and other urinary metabolites were examined in a follow-up study in 15 NEC cases and 30 controls <29 weeks gestational age.

The follow-up study confirmed that the ratio of urinary alanine to histidine is a significant predictor of NEC (Figure 22). In particular, it was found that the alanine-histidine ratio was a predictor of NEC, excluding NEC preceded by sepsis. If sepsis prior to NEC cases are excluded, the alanine-histidine (AH) ratio alone: Predictive value, c=72.5% (Table 12).

Table 12. Association between the ratio of urinary Ala:His and NEC

Logistic regression Number of obs = 42, LR chi2(l) = 9.47, Prob > chi2 = 0.0021

Log likelihood = -20.390146 Pseudo R2 = 0.1885

F. Combination of Urinary Metabolites, high Firmicutes, and urinary glucose

Firmicutes, the alanine-histidine ratio, and urinary glucose were analyzed as a combined marker set for NEC. An 88% predictive value was obtained from the model including detection of high Firmicutes, the alanine-histidine ratio, and the quantity of urinary glucose (xlOOO) when the model includes all NEC cases (Table 13). The predictive value was 90% with exclusion of cases of sepsis prior to NEC. Thus, the strongest prediction occurs with the combination of high Firmicutes measured in stool sample, and an increased urinary alanine-histidine ratio and an increased urinary glucose concentration. Table 13. Association between NEC and the combination of urinary metabolites, high Firmicutes, and urinary glucose

References

1. Stoll, B.J., N.I. Hansen, E.F. Bell, S. Shankaran, A.R. Laptook, et al., Neonatal outcomes of extremely preterm infants from the NICHD Neonatal Research Network.

Pediatrics, 2010, 126:443-56.

2. Smith, P.B., N. Ambalavanan, L. Li, CM. Cotten, M. Laughon, et al., Approach to Infants Born at 22 to 24 Weeks' Gestation: Relationship to Outcomes of More-Mature Infants. Pediatrics, 2012, 129:el508-el516.

3. Yee, W.H., A.S. Soraisham, V.S. Shah, K. Aziz, W. Yoon, et al., Incidence and timing of presentation of necrotizing enterocolitis in preterm infants. Pediatrics, 2012, 129:e298-304.

4. Nanthakumar, N., D. Meng, A.M. Goldstein, W. Zhu, L. Lu, et al., The mechanism of excessive intestinal inflammation in necrotizing enterocolitis: an immature innate immune response. PLoS ONE, 2011, 6:el7776.

5. Meinzen-Derr, J., B. Poindexter, L. Wrage, A.L. Morrow, B. Stoll, et al., Role of human milk in extremely low birth weight infants' risk of necrotizing enterocolitis or death. J Perinatal, 2009, 29:57-62.

6. Quigley, M.A., G. Henderson, M.Y. Anthony, and W. McGuire, Formula milk versus donor breast milk for feeding preterm or low birth weight infants. Cochrane database of systematic reviews (Online), 2007:CD002971. 7. Cotten, CM., S. Taylor, B. Stoll, R.N. Goldberg, N.I. Hansen, et al., Prolonged duration of initial empirical antibiotic treatment is associated with increased rates of necrotizing enterocolitis and death for extremely low birth weight infants. Pediatrics, 2009, 123:58-66.

8. Kuppala, V.S., J. Meinzen-Derr, A.L. Morrow, and K.R. Schibler, Prolonged initial empirical antibiotic treatment is associated with adverse outcomes in premature infants. The Journal of pediatrics, 2011, 159:720-5.

9. Wang, Y., J.D. Hoenig, K.J. Malin, S. Qamar, E.O. Petrof, et al., 16S rRNA gene- based analysis of fecal microbiota from preterm infants with and without necrotizing enterocolitis. ISME J, 2009, 3:944-54.

10. Mai, V., CM. Young, M. Ukhanova, X. Wang, Y. Sun, et al., Fecal microbiota in premature infants prior to necrotizing enterocolitis. PLoS ONE, 2011, 6:e20647.

11. Carlisle, E.M., V. Poroyko, M.S. Caplan, J. A. Alverdy, and D. Liu, Gram negative bacteria are associated with the early stages of necrotizing enterocolitis. PLoS ONE, 2011, 6:el8084.

12. LaTuga, M.S., J.C Ellis, CM. Cotton, R.N. Goldberg, J.L. Wynn, et al., Beyond bacteria: a study of the enteric microbial consortium in extremely low birth weight infants. PLoS ONE, 2011, 6:e27858.

13. Afrazi, A., CP. Sodhi, W. Richardson, M. Neal, M. Good, et al., New insights into the pathogenesis and treatment of necrotizing enterocolitis: Toll-like receptors and beyond.

Pediatric research, 2011, 69: 183-8.

14. Mihatsch, W.A., CP. Braegger, T. Decsi, S. Kolacek, H. Lanzinger, et al., Critical systematic review of the level of evidence for routine use of probiotics for reduction of mortality and prevention of necrotizing enterocolitis and sepsis in preterm infants. Clinical nutrition (Edinburgh, Scotland), 2012, 31:6-15.

15. Morowitz, M.J., V. Poroyko, M. Caplan, J. Alverdy, and D.C Liu, Redefining the role of intestinal microbes in the pathogenesis of necrotizing enterocolitis. Pediatrics, 2010, 125:777-85.

16. Parra-Herran, C.E., L. Pelaez, J.E. Sola, A.K. Urbiztondo, and M.M. Rodriguez, Intestinal candidiasis: an uncommon cause of necrotizing enterocolitis (NEC) in neonates.

Fetal and pediatric pathology, 2010, 29: 172-80. 17. Stuart, R.L., K. Tan, J.E. Mahar, CD. Kirkwood, C. Andrew Ramsden, et al., An outbreak of necrotizing enterocolitis associated with norovirus genotype GII.3. The Pediatric infectious disease journal, 2010, 29:644-7.

18. Tai, I.C., Y.C. Huang, R.I. Lien, C.G. Huang, K.C. Tsao, et al., Clinical

manifestations of a cluster of rotavirus infection in young infants hospitalized in neonatal care units. J Microbiol Immunol Infect, 2012, 45: 15-21.

19. El Aidy, S., P. van Baarlen, M. Derrien, D.J. Lindenbergh-Kortleve, G. Hooiveld, et al., Temporal and spatial interplay of microbiota and intestinal mucosa drive establishment of immune homeostasis in conventionalized mice. Mucosal Immunol, 2012.

20. Hooper, L.V., D.R. Littman, and A.J. Macpherson, Interactions between the microbiota and the immune system. Science (New York, N.Y, 2012, 336: 1268-73.

21. Martin, R., A.J. Nauta, K. Ben Amor, L.M. Knippels, J. Knol, et al., Early life: gut microbiota and immune development in infancy. Benef Microbes, 2010, 1:367-82.

22. Reikvam, D.H., A. Erofeev, A. Sandvik, V. Grcic, F.L. Jahnsen, et al., Depletion of murine intestinal microbiota: effects on gut mucosa and epithelial gene expression. PLoS

ONE, 2011, 6:el7996.

23. Kaplan, J.L., H.N. Shi, and W.A. Walker, The role of microbes in developmental immunologic programming. Pediatric research, 2011, 69:465-72.

24. Mazmanian, S.K., C.H. Liu, A.O. Tzianabos, and D.L. Kasper, An

immunomodulatory molecule of symbiotic bacteria directs maturation of the host immune system. Cell, 2005, 122: 107-18.

25. Cevallos-Cevallos, J.M., M.D. Danyluk, and J.I. Reyes-De-Corcuera, GC-MS based metabolomics for rapid simultaneous detection of Escherichia coli 0157:H7, Salmonella Typhimurium, Salmonella Muenchen, and Salmonella Hartford in ground beef and chicken. Journal of food science, 2011, 76:M238-46.

26. Dai, Z.L., J. Zhang, G. Wu, and W.Y. Zhu, Utilization of amino acids by bacteria from the pig small intestine. Amino acids, 2010, 39: 1201-15.

27. Ponnusamy, K., J.N. Choi, J. Kim, S.Y. Lee, and C.H. Lee, Microbial community and metabolomic comparison of irritable bowel syndrome faeces. Journal of medical

microbiology, 2011, 60:817-27. 28. Swann, J.R., K.M. Tuohy, P. Lindfors, D.T. Brown, G.R. Gibson, et al., Variation in antibiotic-induced microbial recolonization impacts on the host metabolic phenotypes of rats. J Proteome Res, 2011, 10:3590-603.

29. Tuohy, K.M., C. Gougoulias, Q. Shen, G. Walton, F. Fava, et al., Studying the human gut microbiota in the trans-omics era— focus on metagenomics and metabonomics. Curr

Pharm Des, 2009, 15: 1415-27.

30. Walsh, M.C. and R.M. Kliegman, Necrotizing enterocolitis: treatment based on staging criteria. Pediatric clinics of North America, 1986, 33: 179-201.

31. Romick-Rosendale, L.E., H.I. Brunner, M.R. Bennett, R. Mina, S. Nelson, et al., Identification of urinary metabolites that distinguish membranous lupus nephritis from proliferative lupus nephritis and focal segmental glomerulosclerosis. Arthritis research & therapy, 2011, 13:R199.

32. Evaluation of 16S rDNA-based community profiling for human microbiome research. PLoS ONE, 2012, 7:e39315.

33. Schloss, P.D., S.L. Westcott, T. Ryabin, J.R. Hall, M. Hartmann, et al., Introducing mothur: open-source, platform-independent, community-supported software for describing and comparing microbial communities. Applied and environmental microbiology, 2009, 75:7537-41.

34. Edgar, R.C., B.J. Haas, J.C. Clemente, C. Quince, and R. Knight, UCHIME improves sensitivity and speed of chimera detection. Bioinformatics (Oxford, England), 2011, 27:2194-

200.

35. Kostic, A.D., D. Gevers, C.S. Pedamallu, M. Michaud, F. Duke, et al., Genomic analysis identifies association of Fusobacterium with colorectal carcinoma. Genome research, 2012, 22:292-8.

36. Margulies, M., M. Egholm, W.E. Altman, S. Attiya, J.S. Bader, et al., Genome sequencing in microfabricated high-density picolitre reactors. Nature, 2005, 437:376-80. 37. Cole, J.R., Q. Wang, E. Cardenas, J. Fish, B. Chai, et al., The Ribosomal Database Project: improved alignments and new tools for rRNA analysis. Nucleic acids research, 2009, 37:D141-5.

38. McDonald, D., M.N. Price, J. Goodrich, E.P. Nawrocki, T.Z. DeSantis, et al., An improved Greengenes taxonomy with explicit ranks for ecological and evolutionary analyses of bacteria and archaea. ISME J, 2012, 6:610-8. 39. R Core Team. R: A Language and Environment for Statistical Computing. 2012; Available from: http://www.R-project.org.

40. Segata, N., J. Izard, L. Waldron, D. Gevers, L. Miropolsky, et al., Metagenomic biomarker discovery and explanation. Genome Biol, 2011, 12:R60.

41. Caporaso, J.G., J. Kuczynski, J. Stombaugh, K. Bittinger, F.D. Bushman, et al., QIIME allows analysis of high-throughput community sequencing data. Nature methods, 2010, 7:335-6.

42. Lozupone, C, M.E. Lladser, D. Knights, J. Stombaugh, and R. Knight, UniFrac: an effective distance metric for microbial community comparison. ISME J, 2011, 5: 169-72. 43. SAS Institute Inc., S.S.V., Cary, NC SAS/STAT Version 9.3: Cary, NC.

44. Chao, A., Nonparametric Estimation of the Number of Classes in a Population. Scandinavian Journal of Statistics, 1984, 11:265-270.

45. Simpson, E.H., Measurement of diversity. Nature, 1949, 163:688.

46. Oksanen, J., F.G. Blanchet, R. Kindt, P. Legendre, R.B. O'Hara, et al. vegan:

Community Ecology Package. 2011; R package version 1.17- 12.: [Available from:

http://cran.r-project.org/package=vegan.

47. McCune, B. and J.B.w.U. Grace, D.L.) Analysis of ecological communities, M.S. Design, Editor 2002: Gleneden Beach, OR, US.

48. Ward, J.H., Hierarchical Grouping to Optimize an Objective Function. Journal of the American Statistical Association, 1963, 58:236-244.

49. Benjamini, Y. and Y. Hochberg, Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing. Journal of the Royal Statistical Society. Series B (Methodological), 1995, 57:289-300.

50. Hotelling, H., The Generalization of Student's Ratio. The Annals of Mathematical Statistics, 1931, 2:360-378.

51. Koenig, J.E., A. Spor, N. Scalfone, A.D. Fricker, J. Stombaugh, et al., Succession of microbial consortia in the developing infant gut microbiome. Proceedings of the National Academy of Sciences of the United States of America, 2011, 108 Suppl 1:4578-85.

52. Morowitz, M.J., V.J. Denef, E.K. Costello, B.C. Thomas, V. Poroyko, et al., Strain- resolved community genomic analysis of gut microbial colonization in a premature infant.

Proceedings of the National Academy of Sciences of the United States of America, 2011, 108: 1128-33. 53. Yatsunenko, T., F.E. Rey, M.J. Manary, I. Trehan, M.G. Dominguez-Bello, et al., Human gut microbiome viewed across age and geography. Nature, 2012, 486:222-7.

54. Foligne, B., S.M. Deutsch, J. Breton, F.J. Cousin, J. Dewulf, et al., Promising immunomodulatory effects of selected strains of dairy propionibacteria as evidenced in vitro and in vivo. Applied and environmental microbiology, 2010, 76:8259-64.

55. McDowell, A., E. Barnard, I. Nagy, A. Gao, S. Tomida, et al., An Expanded

Multilocus Sequence Typing Scheme for Propionibacterium acnes: Investigation of

'Pathogenic', 'Commensal' and Antibiotic Resistant Strains. PLoS ONE, 2012, 7:e41480.

56. Kanehisa, M., S. Goto, Y. Sato, M. Furumichi, and M. Tanabe, KEGG for integration and interpretation of large-scale molecular data sets. Nucleic acids research, 2012, 40:D109-

14.

57. Taur, Y., J.B. Xavier, L. Lipuma, C. Ubeda, J. Goldberg, et al., Intestinal domination and the risk of bacteremia in patients undergoing allogeneic hematopoietic stem cell transplantation. Clin Infect Dis, 2012, 55:905-14.

58. Duerr, C.U. and M.W. Hornef, The mammalian intestinal epithelium as integral player in the establishment and maintenance of host- microbial homeostasis. Semin Immunol, 2012, 24:25-35.

59. Kurokawa, K., J.H. Gong, K.H. Ryu, L. Zheng, J.H. Chae, et al., Biochemical characterization of evasion from peptidoglycan recognition by Staphylococcus aureus D- alanylated wall teichoic acid in insect innate immunity. Dev Comp Immunol, 2011, 35:835-9.

60. Carrasco-Lopez, C, A. Rojas-Altuve, W. Zhang, D. Hesek, M. Lee, et al., Crystal structures of bacterial peptidoglycan amidase AmpD and an unprecedented activation mechanism. The Journal of biological chemistry, 2011, 286:31714-22.

61. Tanaka, T., N. Kitamoto, X. Jiang, and M.K. Estes, High efficiency cross-reactive monoclonal antibody production by oral immunization with recombinant norwalk virus-like particles. Microbiology and immunology, 2006, 50:883-8.

62. Bouladoux, N., J.A. Hall, J.R. Grainger, L.M. Dos Santos, M.G. Kann, et al.,

Regulatory role of suppressive motifs from commensal DNA. Mucosal Immunol, 2012.

63. Blaser, M.J., M.G. Dominguez-Bello, M. Contreras, M. Magris, G. Hidalgo, et al., Distinct cutaneous bacterial assemblages in a sampling of South American Amerindians and

US residents. ISME J, 2012. 64. Jimenez, E., S. Delgado, L. Fernandez, N. Garcia, M. Albujar, et al., Assessment of the bacterial diversity of human colostrum and screening of staphylococcal and enterococcal populations for potential virulence factors. Research in microbiology, 2008, 159:595-601.

65. Kim, M., S. Christley, J.C. Alverdy, D. Liu, and G. An, Immature oxidative stress management as a unifying principle in the pathogenesis of necrotizing enterocolitis: insights from an agent-based model. Surg Infect (Larchmt), 2012, 13: 18-32.

66. El Aidy, S., P. van Baarlen, M. Derrien, D.J. Lindenbergh-Kortleve, G. Hooiveld, et al., Temporal and spatial interplay of microbiota and intestinal mucosa drive establishment of immune homeostasis in conventionalized mice. Mucosal Immunol, 2012, 5:567-79.

OTHER EMBODIMENTS

All of the features disclosed in this specification may be combined in any

combination. Each feature disclosed in this specification may be replaced by an alternative feature serving the same, equivalent, or similar purpose. Thus, unless expressly stated otherwise, each feature disclosed is only an example of a generic series of equivalent or similar features.

From the above description, one skilled in the art can easily ascertain the essential characteristics of the present invention, and without departing from the spirit and scope thereof, can make various changes and modifications of the invention to adapt it to various usages and conditions. Thus, other embodiments are also within the claims.

Claims

What Is Claimed Is:

1. A method of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the method comprising:

providing a urine sample of a preterm infant;

measuring a level of a metabolite biomarker in the urine sample, wherein the metabolite biomarker is vitamin B₆, a vitamin B₆ metabolite, alanine, histidine, or a combination thereof; and

identifying the preterm infant as having or at risk for NEC if the level of the metabolite biomarker deviates from a reference level.

2. The method of claim 1, wherein the urine sample is obtained from the preterm infant during postnatal day 4 to postnatal day 9.

3. The method of claim 1 or 2, wherein the metabolite biomarker is vitamin B₆ or a vitamin B₆ metabolite.

4. The method of claim 3, wherein the preterm infant is identified as an infant having or at risk for type I NEC if the level of the vitamin B₆ or the vitamin B₆ metabolite in the urine sample is elevated relative to the reference level.

5. The method of claim 1 or 2, wherein the metabolite biomarker is alanine.

6. The method of claim 5, wherein the preterm infant is identified as an infant having or at risk for type I NEC if the level of alanine in the urine sample is elevated relative to the reference level.

7. The method of claim 1 or 2, wherein the metabolite biomarker is a

combination of alanine and histidine, and wherein the method further comprises:

calculating a ratio between the level of alanine and the level of histidine, and identifying the infant as having or at risk for NEC if the ratio deviates from a reference level.

8. The method of claim 7, wherein the infant is identified as having or at risk for NEC if the ratio of the level of alanine to the level of histidine is greater than 4.

9. The method of claim 1 or 2, wherein the metabolite biomarker is histidine.

10. The method of claim 9, wherein the preterm infant is identified as an infant having or at risk for type II NEC if the level of histidine is reduced relative to the reference level.

11. The method of any of claims 1-10, wherein the level of the metabolite biomarker is measured by nuclear magnetic resonance (NMR) spectroscopy.

12. A method of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the method comprising:

providing at least one stool sample of the preterm infant;

determining bacteria diversity of the microbiota in the stool sample; and

identifying the preterm infant as an infant having or at risk for developing NEC based on the diversity of the microbiota.

13. The method of claim 12, wherein the at least one stool sample is obtained from the preterm infant during postnatal day 4 to postnatal day 9.

14. The method of claim 13, wherein the preterm infant is identified as an infant having or at risk for type I NEC if the diversity of the microbiota represents a high relative abundance of Firmicutes.

15. The method of claim 14, wherein the preterm infant is identified as an infant having or at risk for type I NEC if the relative abundance of Firmicutes is greater than 80%.

16. The method of claim 14 or 15, wherein the Firmicutes are Bacilli bacteria.

17. The method of claim 14 or 15, wherein the Firmicutes are Staphylococcaceae bacteria, Enterococcaceae bacteria, or both.

18. The method of claim 17, wherein the Staphylococcaceae bacteria are Staphylococcus bacteria.

19. The method of claim 12, wherein the at least one stool sample is obtained during postnatal day 10 to postnatal day 16.

20. The method of claim 19, wherein the preterm infant is identified as an infant having or at risk for type II NEC if the bacterial diversity of the microbiota represents a high relative abundance of Proteobacteria.

21. The method of claim 20, wherein the preterm infant is identified as an infant having or at risk for type II NEC if the relative abundance of Proteobacteria is greater than

80%.

22. The method of claim 20 or 21, wherein the Proteobacteria are

Enterobacteriaceae bacteria.

23. The method of claim 12, wherein the at least one stool sample includes a first stool sample and a second stool sample, the first stool sample being obtained from the preterm infant during postnatal day 4 to postnatal day 9 and the second stool sample being obtained from the preterm infant during postnatal day 10 to postnatal day 16.

24. The method of claim 23, wherein the preterm infant is identified as an infant having or at risk for NEC if the bacteria diversity of the microbiota in the second stool sample represents a decrease in the relative abundance of Firmicutes, an increase in the relative abundance of Proteobacteria, or both, as compared to the bacteria diversity of the microbiota in the first stool sample.

25. The method of any of claims 12-24, wherein the bacterial diversity is determined by analyzing the 16s RNAs amplified from the stool sample.

26. A method of identifying a preterm infant having or at risk for necrotizing enterocolitis (NEC), the method comprising:

providing at least one stool sample and at least one urine sample of the preterm infant; determinining bacteria diversity of the microbiota in the stool sample;

measuring a level of a metabolite biomarker in the urine sample, wherein the metabolite biomarker is vitamin B₆, a vitamin B₆ metabolite, alanine, histidine, or a combination thereof; and

identifying the preterm infant as an infant having or at risk for developing NEC based on the bacteria diversity and the level of the metabolite.

27. The method of claim 26, wherein both the urine sample and the stool sample are obtained during postnatal day 4 to postnatal day 9.

28. The method of claim 27, wherein the metabolite is alanine, vitamin B₆, a vitamin B₆ metabolite, or a combination thereof.

29. The method of claim 28, wherein the preterm infant is identified as an infant having or at risk for type I NEC if the level of the metabolite is higher than a reference level and the bacteria diversity represents a high relative abundance of Firmicutes.

30. The method of claim 26, wherein the urine sample is obtained during postnatal day 4 to postnatal day 9 and the stool sample is obtained during postnatal day 10 to postnatal day 16.

31. The method of claim 30, wherein the metabolite is histidine.

32. The method of claim 31, wherein the preterm infant is identified as an infant having or at risk for type II NEC if the level of histidine is lower than a reference level and the bacteria diversity represents a high relative abundance of Proteobacteria.

33. The method of any of claims 26-32, wherein the level of the metabolite is determined by NMR.

34. The method of any of claims 26-33, wherein the bacteria diversity is determined by analyzing the 16s RNAs amplified from the stool sample.

35. The method of any of the preceding claims, further comprising subjecting the preterm infant to a treatment of NEC if the preterm infant is identified as having or at risk for NEC.