Changing the paradigm of defining, detecting, and diagnosing NEC: Perspectives on Bell’s stages and biomarkers for NEC

Seminars in Pediatric Surgery 27 (2018) 3–10

Contents lists available at ScienceDirect

Seminars in Pediatric Surgery journal homepage: www.elsevier.com/locate/sempedsurg

Changing the paradigm of defining, detecting, and diagnosing NEC: Perspectives on Bell’s stages and biomarkers for NEC Sheila M. Gephart, PhD, RNa,n, Phillip V. Gordon, MD, PhDb,c, Alexander H. Penn, PhDd, Katherine E. Gregory, PhD, RNe,f,g, Jonathan R. Swanson, MD, MSch, Akhil Maheshwari, MDi, Karl Sylvester, MDj,k,l a

Community and Health Systems Science, The University of Arizona College of Nursing, PO Box 210203, Tucson, Arizona 85721 Pediatrix-Obstetrix Center for Research and Education, Sunrise, Florida Sacred Heart Children’s Hospital, Pensacola, Florida d TriService Research Laboratory, JBSA-Ft., Sam Houston, Texas e Department of Pediatric Newborn Medicine, Brigham and Women’s Hospital, Boston, Massachusetts f Department of Nursing, Brigham and Women’s Hospital, Boston, Massachusetts g Department of Pediatrics, Harvard Medical School, Boston, Massachusetts h Department of Pediatrics, University of Virginia School of Medicine, Charlottesville, Virginia i Department of Pediatrics, Molecular Medicine and Public Health, University of South Florida, Tampa, Florida j Department of Surgery and Pediatrics, Stanford University School of Medicine, Palo Alto, California k Department of Research, Stanford University School of Medicine, Palo Alto, California l Fetal and Pregnancy Health, Lucile Packard Children’s Hospital Stanford, Palo Alto, California b c

a r t i c l e i n fo

Keywords: Necrotizing enterocolitis Clinical predictors Bell’s stages Clinical definition Spontaneous intestinal perforation Biomarkers

a b s t r a c t Better means to diagnose and define necrotizing enterocolitis are needed to guide clinical practice and research. Adequacy of Bell’s staging system for clinical practice and clarity of cases used in NEC clinical datasets has been a topic of controversy for some time. This article provides reasons why a better global definition for NEC is needed and offers a simple alternative bedside definition for preterm NEC called the “Two out of Three” rule. Some argue that biomarkers may fill knowledge gaps and provide greater precision in defining relevant features of a clinical disease like NEC. NEC biomarkers include markers of inflammation, intestinal dysfunction, hematologic changes, and clinical features. Development and reporting of NEC biomarkers should be guided by the FDA’s BEST Consensus resource, “Biomarkers, EndpointS, & other Tools” and consistently report metrics so that studies can be compared and results pooled. Current practice in the NICU would be enhanced by clinical tools that effectively inform the clinical team that a baby is at increasing risk of NEC. Ideally, these tools will incorporate both clinical information about the baby as well as molecular signals that are indicative of NEC. While meaningful biomarkers for NEC and clinical tools exist, translation into practice is mediocre. & 2018 Elsevier Inc. All rights reserved.

Introduction The field of necrotizing enterocolitis (NEC) research and the diagnosis of NEC itself has existed for more than 60 years.1,2 It has preceded and transcended all the different forms of continuous positive ventilation, and it has been transformed in the era of surfactants.3 The field has also seen the emergence of clinical cohort studies (versus case-series), randomized controlled trials, long-term outcome studies and meta-analytics. In short, the field of NEC is older than neonatology itself. For much of that time, the

n

Corresponding author. E-mail address: [email protected] (S.M. Gephart).

https://doi.org/10.1053/j.sempedsurg.2017.11.002 1055-8586/& 2018 Elsevier Inc. All rights reserved.

field has relied on the opinion of the surgeon who created Bell’s criteria to characterize the severity of disease in the pre-surfactant era.4 Bell’s three stages were originally designed to help the bedside clinician determine when surgery was prudent. However, during the age when cohort studies were emerging, Bell’s staging became popular as the means to define NEC cohorts. It has been refined successfully only once5 yet still informs many of the dataset definitions that are in use for billing6 and quality improvement.7 After the advent of surfactant and improved survival of premature newborns, it was determined that the lowest stage contained too many patients with confounding, non-NEC diagnoses. The challenge today lies in the evolution of neonatal patient populations and our ability to refine large datasets, such that these

4

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

populations now represent distinct and often novel patient groups.8,9 Term infants versus preterm infants experience different NEC outcomes and precedent risk factors.10,11 Infants with severe congenital anomalies, even if they have a common endpoint disease like NEC, also have dramatically different presentations and outcomes compared to gestational age-matched cohorts with NEC. Finally, there are non-NEC diseases that lurk within our datasets and literature (such as spontaneous intestinal perforations; SIP), which can fatally sabotage our understanding of NEC.8 Recent success in NEC prevention12 and recognition of NEC subsets (where more stringent definitions are being used),13 demonstrate that we must refine our global NEC definition if we are to progress in our understanding of the disease.14 The success and clinical utility of emerging biomarkers, applied diagnostics like ultrasound and quality initiatives (e.g., bundles and checklists) that aim to reduce NEC are highly dependent on a refined and more precise definition. Each modality promises improved capacity to pinpoint the timing, accuracy, and potentially the sub-cohort categorization of NEC. New technologies and approaches make it imperative that we create a more adaptive definition. Finally, it remains crucial that any new definition be operative at the bedside. Whether in London or Kampala, a clinician must have the ability to make the diagnosis of NEC with reasonable accuracy every time. Promising preventive strategies are emerging, including breast milk feeding, antibiotic stewardship, and probiotics whose success will only be fully realized through dissemination and adoption of more rigorous risk assessment and diagnostic definitions.15–17 As NEC decreases globally we must have the ability to capture confirmed cases in our datasets, correctly, or risk an ever greater pool of non-NEC versus true cases of NEC.14 The purpose of this review is to (1) offer a better definition of NEC based on newer and novel markers and radiographic imaging, and (2) define adequate biomarkers which may be used to help guide clinical decision-making.

Defining NEC Survey of experts about adequacy of current NEC definitions During the 2017 NEC Symposium, participants at a “Defining NEC” workshop were asked questions about the adequacy of current definitions used to define NEC. Prior to the conference, a waiver of consent was obtained from the University of Virginia Social and Behavioral Sciences Institutional Review Board (project #2017-0125-00). Workshop participants were self-selected stakeholders in NEC so initial survey questions were used to identify bias. Respondents to the survey included 20 physicians (4 of whom were surgeons), 18 non-physician clinicians (nurses, neonatal nurse practitioners, dietitians, etc.), 9 non-physician researchers, and 5 others (including one parent). Participants were asked to rank their agreement on a scale of 1–5 (with 5 being strong agreement and 1 being strong disagreement) with statements about ease of NEC diagnosis and reliability of Bell’s staging for creating preterm NEC datasets. Responses are shown in Figure 1. Responses suggest that workshop participants came in with a relatively strong bias against Bell’s staging as a reliable definition and many felt that NEC is not easy to diagnose correctly. Additionally, respondents were asked to relate NEC definition to 17 potential biomarkers, clinical findings, or demographic variables. The participants ranked the order of importance in “forming a correct diagnosis of NEC.” Too few complete responses were available for statistical analysis, potentially signaling a lack of deep understanding and/or a lack of clarity in the evidence about each variable which can be used consistently and reliably to define NEC clinically.

Fig. 1. Pre-workshop perspectives on Bell’s reliability. Responses to two preworkship survey questions. Y-axis ¼ number of respondents.

In the post-workshop survey, participants were asked five questions, with the first four allowing only a single answer choice. Respondents were asked to pick the statement that “best reflects your opinion” regarding the definition of NEC. Results are shown in Figure 2. The majority of participants thought Bell’s staging to be poorly reliable for the diagnosis of preterm NEC. Likewise, the vast majority thought that a new definition of NEC was necessary and that we needed to either abandon or modify Bell’s in order to improve data quality. Finally, the overwhelming majority felt preterm and term NEC were either different diseases entirely or similar but with different preceding events and risks. No survey respondent felt the two entities to be the same disease. A fifth question allowed the participant to check any or all of the responses. It asked, “Which of the following statements best reflects your opinion” on variables used to define NEC. The responses are represented in Figure 3. Respondents picked histopathology and a final common pathway of disease as the two most important variables for defining NEC. Use of subgroups that respond to quality improvement was a respectable third choice. Use of an inclusive definition was not picked often. After workgroups met, leaders presented a summary to all NEC symposium attendees. Dr. Gordon presented a brief synopsis, described the post-workshop responses indicating that we needed to abandon or modify Bell’s staging to improve our data, and led the discussion. There was surprising agreement across a large audience of researchers and clinicians, many with long careers devoted to NEC. When asked to recommend a definition to replace Bell’s staging, Dr. Gordon described the Two out of Three rule, as a place to start for a new bedside definition (see Table 1), which incorporates timing of medical NEC versus SIP onset described elsewhere.18 (Conference attendees suggested that ultrasoundidentified pneumatosis and portal air be considered alternatives to the same radiographic findings and the definition has been modified accordingly.)

Biomarkers for NEC Assessing risk for NEC and diagnostic uncertainty To address the unmet need of reducing or eliminating NEC will require both the development of new tools to assist in defining NEC objectively and a deeper understanding of NEC-associated

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

5

Fig. 2. Post-workshop perspectives on Bell’s reliability and adequacy. Responses to four post-workshop survey questions. Y-axis ¼ number of respondents.

pathophysiology. Diagnostic uncertainty is a challenge in numerous suspected cases of NEC that do not present clinical signs of definitive disease. For example, reliance on pneumatosis or perforation by Bell’s Stage II and III, respectively, typically indicates that the disease is both established and has progressed, potentially limiting the development of novel prevention or therapeutic strategies. Accordingly, it remains entirely unclear if established cases of NEC (≥Bell’s Stage II, or 2 out of 3 per the proposed new definition) can be halted or outcomes improved through the use of alternative therapeutic strategies, e.g., early surgery, or novel therapeutic agents. The use of definitive NEC case definitions and the exclusion of confounding cases of SIP are desirable to achieve greater accuracy in published reporting, however, it is also quite possible that the larger opportunity resides in defining a preclinical state of definitive NEC and identifying those most at risk for NEC. Bell’s is a staging criteria, not a diagnostic criteria, and Bell’s I, wherein the disease is suspected, utilizes highly nonspecific (to NEC) criteria that do not allow one to differentiate NEC from alternative intestinal-specific pathology like feeding intolerance or generalized sepsis with ileus. Moreover, there is a need to move past the well-established age- and weight-based risks for NEC [extremely low birth weight (ELBW, o 1000 g) and very low birth weight ( o1500 g)] and consider what these clinical features indicate biologically and how biologic risk of prematurity is affected by modifiable clinical risk factors (exposures) for NEC; e.g., intestinal colonization, enteral feedings.19 A great many more infants will experience risk factors for NEC than will develop NEC itself. This has restricted us to therapeutic strategies that are supportive rather than preventive, due to the limited utility of prospective studies. To effectively test clinical interventions, we need more selective, sensitive, and objective criteria (clinical, molecular, and imaging) or biomarkers to identify the infants at greatest risk of progressing to NEC prior to

their development of irreversible injury with perforation or gangrene. Standard biomarker definition For consistency, it is beneficial to adopt a standard definition of a biomarker. Accordingly, the NIH in collaboration with the FDA has produced a consensus document, the BEST (Biomarkers, EndpointS, & other Tools) Resource that provides extensive guidance on the various types and utility of biomarkers (e.g., diagnostic, prognostic, predictive, surrogate endpoint, etc.). (https://www. ncbi.nlm.nih.gov/books/NBK338448/) As defined by BEST, a biomarker is “A defined characteristic that is measured as an indicator of normal biological processes, pathogenic processes, or response to an exposure or intervention. Molecular, histologic, radiographic, or physiologic characteristics are types of biomarkers.” Enhancing clinical utility through reporting predictive performance Many published NEC biomarker studies have documented acceptable if not excellent sensitivity (i.e., how likely is the test to detect the presence of a disease in someone with the disease) and specificity (how likely is the test to detect the absence of a disease in someone without the disease).20 This is typically presented as area Table 1 The two out of three rule for bedside diagnosis of preterm NEC.14 Patients may be given a diagnosis of Preterm NEC if they have abdominal distension, ileus and/or bloody stools and meet at least 2 of the criteria below: 1. Pneumatosis and/or portal air by ultrasound or abdominal x-ray at presentation 2. Persistent platelet consumption ( o150,000 × 3 days after diagnosis) 3. Post-menstrual age at disease onset is more consistent with NEC than spontaneous intestinal perforation (SIP)* Patients excluded from a diagnosis of Preterm NEC: 1. Infants known to have SIP 2. Infants with complex congenital anomalies 3. Infants being fed o80 ml/kg/day 4. Infants Z36 weeks gestation

Fig. 3. Priorities to define NEC. The X-axis corresponds to number of respondents.

n See published figures describing timing differences for NEC versus SIP.18 Preterm is defined in this instance as o36 weeks gestational age at birth.

6

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

under the curve (AUC) or a c-statistic on a receiver operator curve (ROC). However, without a clear understanding of disease prevalence (i.e., prior probability) in the population of interest, a highly sensitive test can be very misleading. The clinical utility of diagnostic tests for NEC resides with the predictive value that is highly dependent on disease prevalence, which for a rare disease like NEC (e.g., for VLBW infants, 3–7%),21–23 is more likely to produce a high false positive rate in order to achieve high sensitivity. Contrast the definition of sensitivity provided above, with the Positive Predictive Value (PPV) that indicates how likely is someone with a positive test result to actually have the disease [Positive Predictive Value ¼ True Positive/(True Positive þ False Positive)]. Thus, it is imperative that diagnostic and predictive models and biomarkers of NEC include reports of PPV, and Negative Predictive Value (NPV). A modest PPV may be acceptable for a low-risk intervention, but less so if the proposed intervention risk is high and the False Positive rate is high, resulting in a number of exposed patients (false positives) that could suffer collectively more harm than the small group who benefit (true positives). The consideration of clinical utility nicely illustrates the importance of examining prior probabilities, likelihood prior to test (indicated by disease prevalence) and the impact of observations (exposures) that will affect the test performance in determining a positive predictive value. As can be appreciated, for the development of diagnostic tests, a comparison to a gold standard is required, which for NEC at this time remains confirmed cases (≥Bell’s II) indicated by pneumatosis or necrotic bowel. This gold standard is specific for established or progressive disease, but insensitive to early disease and does not distinguish between NEC subsets. Additional considerations in the development of biomarkers should include comparisons to existing disease biomarkers and diagnostics (e.g., CRP, complete blood cell count, abdominal radiograph, or ultrasound), and the development of composite models utilizing available clinical features together with novel molecular indicators to resolve diagnostic ambiguity.24 Multivariate models with standard measures are critical and Bayesian models that provide conditional probabilities may provide key differential insights.

Biomarkers in NEC The complete blood count has promise as a biomarker of early NEC. An acute drop in absolute monocyte counts,25 shifts

in immature to total neutrophil ratios, drops in platelets and rising eosinophilia after blood transfusion26 are all associated with NEC onset. The degree of platelet consumption relates to NEC severity and is associated with higher mortality.27 Many published reports have utilized existing markers of inflammation (e.g., CRP, various interleukins, α1-antitrypsin, and TGF-β),27,28 and in some cases markers that appear to be more specific for intestinal and NEC pathophysiology (e.g., iFABP and PAF, respectively).29 A recent meta-analysis of 14 studies evaluating performance of iFABP in plasma showed overall fair sensitivity (pooled 0.64, 95% CI: 0.53– 0.74), and strong specificity (pooled 0.91, 95% CI: 0.84–0.95).29 Calprotectin, an inflammatory protein that is expressed and secreted by neutrophils is measured in stool and has been shown to indicate intestinal damage in NEC.30–32 In a 2016 systematic review, Pergialiotis et al.33 showed that 10 of 13 studies revealed elevated fecal calprotectin levels in those who got NEC but the sensitivity and specificity was highly variable and the cut-off values were different across studies. A key challenge in studies of NEC biomarkers is the variability in case and control definitions (comparator groups), study size, and statistical performance (Figure 4). For example, choosing a non-ill control (comparator group) in determining the diagnostic performance of a test for “early NEC” may lead to exaggerated test performance. If the objective is to develop a diagnostic for early NEC, then a more appropriate design would be a possible or established sepsis group from which early or “suspicious NEC” can be re-classified.34 These limitations together with a general absence of prospective validation in multi-center studies to assess analytical validity (reproducibility), clinical validity and generalizability, render most NEC biomarkers published to date of limited clinical utility. Biomarkers studied to date are summarized in Table 2.

Cytokines as biomarkers of NEC Several studies have shown increased tissue expression of interleukin (IL)-1B, IL-8/CXC-motif ligand (CXCL)-8, and tumor necrosis factor (TNF) in surgically resected tissue specimens of NEC.35–37 MohanKumar et al.38 have recently provided a more comprehensive cytokine expression profile in intestinal tissue affected by NEC. Comparing microarray datasets from surgically resected specimens NEC versus uninflamed neonatal

Fig. 4. Statistical performance of biomarkers for NEC. Plot demonstrating variability in sensitivity, specificity, and study design for three NEC biomarker studies. Each circle represents an analyte specific study, and the size of the circle represents relative size (number of participants) of study. Study A (white): NEC compared to a control cohort without disease; Study B (red): NEC and sepsis together compared to a control cohort without disease. Study C (blue): early NEC Stage I compared to late Stage III NEC. Biomarkers displayed include: alkaline phosphatase (ALP);calprotectin; claudin 3; C-reactive protein (CRP); endotoxin units; intestinal fatty acid binding protein (IFABP); intestinal fatty acid binding protein/creatinine (IFABP/Cr); intestinal fatty acid binding protein/serum amyloid A (IFABP/SAA); procalcitonin (PCT); platelet activating factor (PAF); Pro-apoC2/serum amyloid A (Pro-apoC2/SAA); S100 calcium binding protein A12 (S100A12); interleukin-1 receptor agonist (IL-1ra); serum amyloid A (SAA); tumor necrosis factor- alpha (TNF-á) (Color version of figure is available online.)

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

7

Table 2 Biomarkers evaluated in NEC. Biomarker/ selected references

What it is

How it is used

Advantages

i-FABP29

Protein specific to intestinal epithelium

Measured in blood or urine to detect intestinal epithelial cell death

Tested in numerous studies; Overall high specificity, but only moderate specific to intestinal damage; sensitivity can be measured non-invasively in urine Commonly measured; correlates to Not specific to the intestine; may not appear severity and mortality early enough to be a predictive biomarker

Blood platelet count27

Platelets are fragments of megakaryocytes that circulate in blood and are responsible for blood clotting C-reactive Acute-phase plasma protein protein elevated in response to IL-6 (CRP)28,47 and other inflammatory signals Blood monocyte Monocytes are leukocytes that count25 extravasate and differentiate into macrophages in response to inflammatory signals Interleukin-6 Inflammatory cytokine (IL-6)28,39,47

Measured with blood CBC to indicate intestinal gangrene (and possibly disseminated intravascular coagulation) Measured in blood to detect inflammation

Commonly measured; correlates to Not specific to the intestine; may not appear severity early enough to be a predictive biomarker

Measured with blood CBC to indicate monocyte extravasation

Commonly measured; differentiates NEC from other causes of feeding intolerance

Not specific to the intestine; few studies to date

Measured in blood to detect inflammation

Commonly measured cytokine; likely able to distinguish NEC from SIP

Measured in blood to detect inflammation

Sustained elevation; correlates to severity and mortality; likely able to distinguish NEC total is from NEC and NEC from SIP Much like gestational age and weight, TGF-β can indicate infants at greater risk of NEC from time of birth

Not specific to the intestine; may not appear early enough to be a predictive biomarker; may not stay elevated long enough to be a good marker for severity; does not distinguish between NEC and sepsis Not specific to the intestine; may not appear early enough to be a predictive biomarker

Interleukin-8 (IL-8)28,47

Inflammatory cytokine

Transforming growth factor-beta (TGF-β)28,49

Growth factor deficient in the Measured in blood to indicate intestine of premature infants deficits in intestinal development

Calprotectin31,33 Inflammatory protein expressed Measured in feces to detect and secreted by neutrophils neutrophil infiltration of the intestine Serum amyloid Acute-phase plasma protein Measured in blood or urine to 32 A (SAA) detect inflammation elevated in response to IL-6 and other inflammatory signals Platelet activating factor (PAF)36

Phospholipid inflammatory mediator produced by platelets, leukocytes, and endothelial cells

Limitations

Measured in blood to indicate inflammation

intestine resected for intestinal obstruction/during ileostomy repair, they showed that NEC was associated with increased IL1A, IL-1B, IL-6, IL-10, TNF, CXCL1, CXCL2, CXCL3, CXCL5, IL-8/CXCL8, CXCL12, CCL2, CCL3, CCL4, CCL7, CCL8, CCL20, hepatocyte growth factor (HGF), leukemia inhibitory factor (LIF), thymic stromal lymphopoietin (TSLP), platelet factor 4, tumor necrosis factor superfamily member 9 (TNFSF9), transforming growth factor (TGF)-B1, TGFB3, and vascular endothelial growth factor (VEGF)A. NEC downregulated TNFSF10/TNF-related apoptosis-inducing ligand (TRAIL), and C-motif ligand 2 (XCL2). There was a trend toward decreased EGF expression that did not reach statistical significance. Obtaining neonatal intestinal tissue is highly invasive however, making these assays less than ideal candidates for biomarkers. However, a number of studies have also measured plasma concentrations of cytokines from patients with NEC. Maheshwari et al.28 compared serial blood spot cytokine measurements from 104 extremely low-birth-weight infants with NEC versus 893 controls, and showed that NEC onset was associated with elevated IL-6, IL-8/CXCL8, IL-10, IL-18, CCL2, CCL3, neurotrophin-4, and C-reactive protein. Several other studies have focused on serum IL-6, IL-8/CXCL8, IL-10, and IL-18 concentrations during NEC.35,39–44 IL-6 has a short half-life and may be detectable only for short periods after onset of NEC,45 whereas IL-8/CXCL8 levels show

Not specific to the intestine; may be decreased before the environmental components leading to NEC are activated, so is not a definitive indicator (moderate sensitivity and specificity) Concentration is higher in breast fed versus Tested in numerous studies; formula fed infants; no agreed upon specific to intestinal damage; threshold. Few prospective studies to date measured non-invasively Can be measured non-invasively in Not specific to the intestine; may not appear urine; correlates well to severity early enough to be a predictive biomarker; shown to be less useful than fecal calprotectin in improving the predictive power of I-FABP Elevated versus controls at Bell’s Not specific to the intestine; few studies to Stage II in humans date; may be elevated early, but not lead to NEC

sustained elevation40 and are higher in infants with extensive tissue injury and surgical NEC.39,43,46–48 Cytokines are impressive diagnostic markers when infants with NEC are compared to those with feeding intolerance related to prematurity itself or other noninfectious causes.39,43,46–48 However, the diagnostic accuracy becomes less impressive when compared with infants who have sepsis-related ileus.39,43,46–48 ELBW infants who develop NEC may have lower circulating TGF-β1 levels than controls since the time of birth and prior to the onset of NEC.28 The diagnostic accuracy of circulating TGF-β1 levels seems to be modest. Specifically, blood TGF-β1 concentrations o1380 pg/mL on the first postnatal day predicted future onset of NEC with 64% accuracy. However, even though genetic factors may not be critical to NEC pathogenesis,28 the ability of blood TGF-β1 to estimate the risk of NEC on day 1 is surprising and indicates a need for further study to identify the role of intrauterine and genetic factors.27 These findings were consistent with earlier reports that preterm infants at risk of NEC also have low TGF-β bioactivity in the intestine.49 The reasons for low blood TGF-β1 levels prior to the onset of NEC are unclear; the authors speculated about increased peripheral uptake as a mechanism for low circulating levels in infants with an accentuated deficiency of TGF-β in the intestine due to mucosal inflammation or other genetic/epigenetic factors.50 Regardless, low TGF-β1 level at birth could prove useful as an

8

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

inclusion criteria for prospective studies of therapies and preventative measures. Prognostic clinical tools Prognostic studies highlight low birth weight and gestational age, ethnicity, presence of sepsis, hypotension requiring vasopressor support, and outborn status as high to moderate quality clinical risk factors for NEC.51 Need for assisted ventilation, birth via Cesarean section, premature rupture of membranes, small for gestational age, and surfactant administration have been identified as moderate quality factors determining risk for NEC.51 All of these factors are confounded by severity of illness, making it difficult to discern whether the risk factor is truly prognostic for NEC or simply associated with increasing severity of illness and in turn, NEC onset. Promising tools that weight and combine these clinical risk factors include the GutCheckNEC score,23 eNEC,52 and the NeoNEEDS measure of intestinal dysfunction.53 Researchers need to move from identifying risk profiles to creating testable, weighted scores that can be compared head-to-head and measured for impact on care decisions. One additional use for such clinical risk tools is to identify infants at highest risk to enroll in clinical trials testing therapeutics or biomarkers. Figure 5 depicts a synthesis of prospective risk factors for NEC and clinical signs that preceded a NEC diagnosis.25,26,58–60

Implications for future work To move the field forward, both precise definitions for NEC and clear criteria for conducting and reporting studies for NEC prognostics (including biomarkers) are needed. We have presented one alternative to Bell’s staging called the “Two out of Three” rule that incorporates items available to the neonatologist and pediatric surgeon to distinguish between preterm NEC and other confounding types of acquired neonatal intestinal diseases (e.g., term NEC and SIP). One of the outcomes of the NEC symposium was the formation of a task force to work in conjunction with the

International Neonatal Consortium of the Critical Path Institute to develop a definition of NEC that would be widely utilized and white papers are forthcoming. Improving the precision of NEC definitions will enable clean clinical datasets and support testing of effective predictive and diagnostic biomarkers. Reporting of biomarker studies should be designed to support meta-synthesis of findings as Yang et al. demonstrated. However, consistent reporting of the pre-test likelihood, positive predictive value and negative predictive value of biomarkers is needed to maximize clinical utility and comparability.54–56 Design and reporting of prognostic studies should follow the recently released “Transparent Reporting of a multivariable prediction model for Individual Prognosis Or Diagnosis (TRIPOD)” statement for scores or algorithms57 and the BEST statement for biomarkers. The next generation of predictive clinical tools must harness a bioinformatic approach including clinical and laboratory data from the electronic health record systems to quickly identify early NEC.61 While meaningful biomarkers for NEC and clinical tools exist, it is unlikely that a single biomarker will be identified and translated as a stand-alone diagnostic test with rapid turnaround suitable for clinical practice. In one published study, fecal calprotectin has been shown to be measurable by a bedside point-of-care device. The measures obtained using this device were well correlated to those obtained using ELISA. While the findings from this study are promising, several steps including those required by the FDA for biomarker identification and validation, guided by the BEST statement, are required before the test or device is widely available at the bedside. Determining the predictive value of combining changes in the complete blood count with plasma iFABP and/or fecal calprotectin would be valuable. Current practice in the NICU would be enhanced by clinical tools that effectively inform the clinical team that a baby is at increasing risk of NEC. Ideally, these tools will incorporate both clinical information about the baby as well as molecular signals that are indicative of NEC. This latter approach would make use of large datasets and advanced analytical approaches such as machine learning, resulting in a personalized medicine approach

Fig. 5. Clinical symptoms antecedent to NEC. Abdominal signs and symptoms antecedent to NEC beginning 36 hours prior to diagnosis. In Stage III NEC requiring surgery, systemic signs and symptoms often precede abdominal signs and symptoms, as indicated by increased apnea and bradycardia, skin mottling, and irritability when compared to increased abdominal girth, blood in stool, and increased pre-feed gastric residual or emesis.

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

to risk stratification among preterm infants for NEC, and better selection of research subjects in studies of therapies to prevent or treat NEC.

Acknowledgment Dr. Gephart acknowledges support from the Robert Wood Johnson Foundation Nurse Faculty Scholars Program (72114) and the Agency for Healthcare Research and Quality (K08HS022908). The content is solely the responsibility of the authors and does not necessarily represent the official views of the Agency for Healthcare Research and Quality or Robert Wood Johnson Foundation. The Necrotizing Enterocolitis Symposium: A Transdisciplinary Approach to Improved NEC Outcomes was supported by a Eugene Washington PCORI Engagement Award by the Patient-Centered Outcomes Research Institute (PCORI) and the University of California Davis. PCORI is an independent, non-profit organization authorized by Congress in 2010 to fund comparative effectiveness research that will provide patients, their caregivers, and clinicians with the evidence needed to make better-informed health and healthcare decisions. PCORI is committed to seeking input from a broad range of stakeholders to guide its work. Dr. Maheshwari is supported by the NIH awards HL124078 and HL133022. We thank Dr. Robert Christensen for his contributions to the workgroup on defining NEC. References 1. Rabl R. Necrotizing enterocolitis in premature infants. Beitr Pathol Anat. 1957;117(2):266–282. 2. Eroess A, Loerinczi K, Nemeth N. Enterocolitis necroticans in newborn infants. Kinderarztl Prax. 1959;27:403–406. 3. Grosfeld JL, Cheu H, Schlatter M, West KW, Rescorla FJ. Changing trends in necrotizing enterocolitis: experience with 302 cases in two decades. Ann Surg. 1991;214(3):300–306. 4. Bell MJ, Ternberg JL, Feigin RD, et al. Neonatal necrotizing enterocolitis. therapeutic decisions based upon clinical staging. Ann Surg. 1978;187(1):1–7. 5. Walsh MC, Kliegman RM. Necrotizing enterocolitis: treatment based on staging criteria. Pediatr Clin North Am. 1986;33(1):179–201. 6. International Classification of Diseases (ICD)-10 2016 version. 2016. http:// www.who.int/classifications/icd/icdonlineversions/en/. 7. Vermont Oxford Network. 2018 Manual of Operations: Data Definitions and Infant Data Forms v. 22.0. In: Vermont2017, Burlington, https://public.vtoxford. org/wp-content/uploads/2017/04/Manual_of_Operations_Part2_v22-1.pdf. Accessed 23.06.2017. 8. Gordon PV, Swanson JR, Attridge JT, Clark R. Emerging trends in acquired neonatal intestinal disease: is it time to abandon Bell’s criteria? J Perinatol. 2007;27(11):661–671. 9. Gordon PV, Swanson JR. Necrotizing enterocolitis is one disease with many origins and potential means of prevention. Pathophysiology. 2014;21(1):13–19. 10. Ostlie DJ, Spilde TL, St Peter SD, et al. Necrotizing enterocolitis in full-term infants. J Pediatr Surg. 2003;38(7):1039–1042. 11. Christensen RD, Lambert DK, Baer VL, Gordon PV. Necrotizing enterocolitis in term infants. Clin Perinatol. 2013;40(1):69–78. 12. Patel AL, Trivedi S, Bhandari NP, et al. Reducing necrotizing enterocolitis in very low birth weight infants using quality-improvement methods. J Perinatol. 2014;34(11):850–857. 13. Gordon P, Christensen R, Weitkamp JH, Maheshwari A. Mapping the new world of necrotizing enterocolitis (NEC): review and opinion. EJ Neonatol Res. 2012; 2(4):145–172. 14. Gordon PV, Swanson JR, MacQueen BC, Christensen RD. A critical question for NEC researchers: can we create a consensus definition of NEC that facilitates research progress? Semin Perinatol. 2017;41(1):7–14. 15. Lee HC, Kurtin PS, Wight NE, et al. A quality improvement project to increase breast milk use in very low birth weight infants. Pediatrics. 2012;130(6): e1679–1687. 16. Benjamin J, Chong E, Reynolds J, Gordon PV, Smith JR. Detailed analysis of NEC risks across a decade in a low incidence NICU : can we drive the incidence of NEC toward zero ? Early enteral feeding for the very low birth weight infant: the development and impact of a research-based guideline. EJ Neonatol Res. 2012;2(4):181–190. 17. Cacho NT, Parker LA, Neu J. Necrotizing enterocolitis and human milk feeding: a systematic review. Clin Perinatol. 2017;44(1):49–67. 18. Gordon PV, Clark R, Swanson JR, Spitzer A. Can a national dataset generate a nomogram for necrotizing enterocolitis onset? J Perinatol. 2014;34(10): 732–735.

9

19. Sylvester KG, Kastenberg ZJ, Moss RL, et al. Acylcarnitine profiles reflect metabolic vulnerability for necrotizing enterocolitis in newborns born premature. J Pediatr. 2017;181(80-85):e81. 20. Evennett N, Alexander N, Petrov M, Pierro A, Eaton S. A systematic review of serologic tests in the diagnosis of necrotizing enterocolitis. J Pediatr Surg. 2009;44(11):2192–2201. 21. Horbar JD, Carpenter JH, Badger GJ, et al. Mortality and neonatal morbidity among infants 501 to 1500 grams from 2000 to 2009. Pediatrics. 2012;129(6): 1019–1026. 22. Yee WH, Soraisham AS, Shah VS, Aziz K, Yoon W, Lee SK. Incidence and timing of presentation of necrotizing enterocolitis in preterm infants. Pediatrics. 2012;129(2):e298–304. 23. Gephart SM, Spitzer AR, Effken JA, Dodd E, Halpern M, McGrath JM. Discrimination of GutCheckNEC: a clinical risk index for necrotizing enterocolitis. J Perinatol. 2014;34(6):468–475. 24. Sylvester KG, Ling XB, Liu GY, et al. A novel urine peptide biomarker-based algorithm for the prognosis of necrotising enterocolitis in human infants. Gut. 2014;63(8):1284–1292. 25. Remon J, Kampanatkosol R, Kaul RR, Muraskas JK, Christensen RD, Maheshwari A. Acute drop in blood monocyte count differentiates NEC from other causes of feeding intolerance. J Perinatol. 2014;34(7):549–554. 26. Christensen RD, Lambert DK, Gordon PV, Baer VL, Gerday E, Henry E. Neonates presenting with bloody stools and eosinophilia can progress to two different types of necrotizing enterocolitis. J Perinatol. 2012;32(11):874–879. 27. Maheshwari A. Immunologic and hematological abnormalities in necrotizing enterocolitis. Clin Perinatol. 2015;42(3):567–585. 28. Maheshwari A, Schelonka RL, Dimmitt RA, et al. Cytokines associated with necrotizing enterocolitis in extremely-low-birth-weight infants. Pediatr Res. 2014;76(1):100–108. 29. Yang G, Wang Y, Jiang X. Diagnostic value of intestinal fatty-acid-binding protein in necrotizing enterocolitis: a systematic review and meta-analysis. Indian J Pediatr. 2016;83(12-13):1410–1419. 30. Bin-Nun A, Booms C, Sabag N, Mevorach R, Algur N, Hammerman C. Rapid fecal calprotectin (FC) analysis: point of care testing for diagnosing early necrotizing enterocolitis. Am J Perinatol. 2015;32(4):337–342. 31. MacQueen BC, Christensen RD, Yost CC, et al. Elevated fecal calprotectin levels during necrotizing enterocolitis are associated with activated neutrophils extruding neutrophil extracellular traps. J Perinatol. 2016;36(10):862–869. 32. Reisinger KW, Kramer BW, Van der Zee DC, et al. Non-invasive serum amyloid A (SAA) measurement and plasma platelets for accurate prediction of surgical intervention in severe necrotizing enterocolitis (NEC). PLoS One. 2014;9(6): e90834. 33. Pergialiotis V, Konstantopoulos P, Karampetsou N, et al. Calprotectin levels in necrotizing enterocolitis: a systematic review of the literature. Inflamm Res. 2016;65(11):847–852. 34. Sylvester KG, Ling XB, Liu GY, et al. Urine protein biomarkers for the diagnosis and prognosis of necrotizing enterocolitis in infants. J Pediatr. 2014;164(3): 607–612 e601–607. 35. Viscardi RM, Lyon NH, Sun CC, Hebel JR, Hasday JD. Inflammatory cytokine mRNAs in surgical specimens of necrotizing enterocolitis and normal newborn intestine. Pediatr Pathol Lab Med. 1997;17(4):547–559. 36. Caplan MS, Sun XM, Hseuh W, Hageman JR. Role of platelet activating factor and tumor necrosis factor-alpha in neonatal necrotizing enterocolitis. J Pediatr. 1990;116(6):960–964. 37. Tan X, Hsueh W, Gonzalez-Crussi F. Cellular localization of tumor necrosis factor (TNF)-alpha transcripts in normal bowel and in necrotizing enterocolitis. TNF gene expression by paneth cells, intestinal eosinophils, and macrophages. Am J Pathol. 1993;142(6):1858–1865. 38. MohanKumar K, Namachivayam K, Ho TT, Torres BA, Ohls RK, Maheshwari A. Cytokines and growth factors in the developing intestine and during necrotizing enterocolitis. Semin Perinatol. 2016;41(1):52–60. 39. Harris MC, Costarino AT Jr., Sullivan JS, et al. Cytokine elevations in critically ill infants with sepsis and necrotizing enterocolitis. J Pediatr. 1994;124(1): 105–111. 40. Edelson MB, Bagwell CE, Rozycki HJ. Circulating pro- and counterinflammatory cytokine levels and severity in necrotizing enterocolitis. Pediatrics. 1999;103(4 Pt 1):766–771. 41. Ford HR, Sorrells DL, Knisely AS. Inflammatory cytokines, nitric oxide, and necrotizing enterocolitis. Semin Pediatr Surg. 1996;5(3):155–159. 42. Ng PC, Li K, Wong RP, et al. Proinflammatory and anti-inflammatory cytokine responses in preterm infants with systemic infections. Arch Dis Child Fetal Neonatal Ed. 2003;88(3):F209–213. 43. Harris MC, D’Angio CT, Gallagher PR, Kaufman D, Evans J, Kilpatrick L. Cytokine elaboration in critically ill infants with bacterial sepsis, necrotizing entercolitis, or sepsis syndrome: correlation with clinical parameters of inflammation and mortality. J Pediatr. 2005;147(4):462–468. 44. Benkoe T, Baumann S, Weninger M, et al. Comprehensive evaluation of 11 cytokines in premature infants with surgical necrotizing enterocolitis. PloS One. 2013;8(3):e58720. 45. Lodha A, Asztalos E, Moore AM. Cytokine levels in neonatal necrotizing enterocolitis and long-term growth and neurodevelopment. Acta Paediatr. 2010;99(3):338–343. 46. Benkoe T, Reck C, Pones M, et al. Interleukin-8 predicts 60-day mortality in premature infants with necrotizing enterocolitis. J Pediatr Surg. 2014;49(3): 385–389.

10

S.M. Gephart et al. / Seminars in Pediatric Surgery 27 (2018) 3–10

47. Bhatia AM, Stoll BJ, Cismowski MJ, Hamrick SE. Cytokine levels in the preterm infant with neonatal intestinal injury. Am J Perinatol. 2014;31(6): 489–496. 48. Sharma R, Tepas JJ 3rd, Hudak ML, et al. Neonatal gut barrier and multiple organ failure: role of endotoxin and proinflammatory cytokines in sepsis and necrotizing enterocolitis. J Pediatr Surg. 2007;42(3):454–461. 49. Maheshwari A, Kelly DR, Nicola T, et al. TGF-β2 suppresses macrophage cytokine production and mucosal inflammatory responses in the developing intestine. Gastroenterology. 2011;140(1):242–253. 50. Namachivayam K, Blanco CL, MohanKumar K, et al. Smad7 inhibits autocrine expression of TGF-beta2 in intestinal epithelial cells in baboon necrotizing enterocolitis. Am J Physiol Gastrointest Liver Physiol. 2013;304(2):G167–180. 51. Samuels N, van de Graaf RA, de Jonge RCJ, Reiss IKM, Vermeulen MJ. Risk factors for necrotizing enterocolitis in neonates: a systematic review of prognostic studies. BMC Pediatr. 2017;17(1):105. 52. Naberhuis J, Wetzel C, Tappenden KA. A Novel neonatal feeding intolerance and necrotizing enterocolitis risk-scoring tool is easy to use and valued by nursing staff. Adv Neonatal Care. 2016;16(3):239–244. 53. Fox JR, Thacker LR, Hendricks-Muñoz KD. Early detection tool of intestinal dysfunction: impact on necrotizing enterocolitis severity. Am J Perinatol. 2015;32(10):927–932.

54. Moons KG, Altman DG, Vergouwe Y, Royston P. Prognosis and prognostic research: application and impact of prognostic models in clinical practice. Br Med J. 2009;338:b606. 55. Hemingway H, Croft P, Perel P, et al. Prognosis research strategy (PROGRESS) 1: a framework for researching clinical outcomes. Br Med J. 2013;346:e5595. 56. Steyerberg EW, Moons KG, van der Windt DA, et al. Prognosis Research Strategy (PROGRESS) 3: prognostic model research. PLoS Med. 2013;10(2):e1001381. 57. Moons KG, Altman DG, Reitsma JB, Collins GS. New guideline for the reporting of studies developing, validating, or updating a multivariable clinical prediction model: the TRIPOD statement. Adv Anat Pathol. 2015;22(5):303–305. 58. Christensen RD, Wiedmeier SE, Baer VL, et al. Antecedents of Bell stage III necrotizing enterocolitis. J Perinatol. 2010;30(1):54–57. 59. Gregory KE, Deforge CE, Natale KM, Phillips M, Van Marter LJ. Necrotizing enterocolitis in the premature infant: neonatal nursing assessment, disease pathogenesis, and clinical presentation. Adv Neonatal Care. 2011;11(3):155–164 quiz 165-6. 60. Gephart SM, Fleiner M, Kijewski A. The conNECtion between abdominal signs and necrotizing enterocolitis in infants 501 to 1500 g. Adv Neonatal Care. 2017;17(1):53–64. 61. Ji J, Ling XB, Zhao Y, et al. A data-driven algorithm integrating clinical and laboratory features for the diagnosis and prognosis of necrotizing enterocolitis. PLoS One. 2014;9(2):e89860.