Article Text
Abstract
Biomarkers have been used to differentiate systemic neonatal infection and necrotising enterocolitis (NEC) from other non-infective neonatal conditions that share similar clinical features. With increasing understanding in biochemical characteristics of different categories of biomarkers, a specific mediator or a panel of mediators have been used in different aspects of clinical management in neonatal sepsis/NEC. This review focuses on how these biomarkers can be used in real-life clinical settings for daily surveillance, bedside point-of-care testing, early diagnosis and predicting the severity and prognosis of neonatal sepsis/NEC. In addition, with recent development of ‘multi-omic’ approaches and rapid advancement in knowledge of bioinformatics, more novel biomarkers and unique signatures of mediators would be discovered for diagnosis of specific diseases and organ injuries.
- biomarkers
- infants
- infection
- necrotising enterocolitis
Statistics from Altmetric.com
Introduction
Systemic neonatal infection and necrotising enterocolitis (NEC) are devastating complications of prematurity. NEC causes intense inflammation of the bowel and often coexists with sepsis. In both conditions, the initial signs are vague and non-specific, and clinically, there is much difficulty in differentiating genuine infection/NEC from other non-infective causes, for example, exacerbation of bronchopulmonary dysplasia or gastrointestinal (GI) dysmotility of prematurity.1 ,2 Laboratory biomarkers that can assist in accurate diagnosis or predicting the severity and prognosis of sepsis/NEC will definitely be useful for clinical management.1 ,2 In addition, novel biomarkers should be discovered: for daily screening of these conditions so that neonatologists may be able to identify them ‘before’ clinical presentation; as a rapid, bedside point-of-care test so as to avoid unnecessary use of antibiotics; and for diagnosing a specific disease or for assessing the extent of organ injury (ie, as a disease-specific or organ-specific biomarker). This review focuses on principles governing the use of laboratory diagnostic biomarkers and how different biomarkers could be used in real-life clinical settings.
The ‘ideal biomarker’ of neonatal infection/NEC
We have in our previous article summarised the critical clinical and laboratory properties of the ‘ideal biomarker’ of neonatal infection.1 Although the basic principles defining an ‘ideal biomarker’ remain unchanged, new practical criteria have been added to the updated table. First, it is of prime importance that infants with systemic neonatal infection/NEC are identified early and accurately differentiated from non-sepsis/non-NEC cases.1 Daily surveillance offers the best hope for detecting these devastating conditions early and before clinical manifestations (box 1, point A2).3–5 For this purpose, the volume of blood required should be minute (<0.1 mL) so that it can be collected for daily screening, and there should be excellent agreement of the biomarker levels between capillary and venous samples in order that life-saving venous access of the infant can be preserved (box 1, points B2–B3).3 Second, an ideal biomarker should be able to help neonatologists monitor the progress of treatment and development of complications, for example, intra-abdominal abscess in NEC (box 1, point A6),6 plus assessing the severity of illness as well as predicting outcomes.7 ,8 Third, an ideal biomarker should be able to accurately diagnose a specific disease or identify specific organ injury. Hence, it should have the ability to differentiate closely related disorders, such as neonatal sepsis from NEC with concurrent bacterial septicaemia (box 1, point A5).9 Thus far, no single biomarker can fulfil all the criteria of an ideal biomarker (box 1), but newer generations of biomarkers are becoming more sophisticated and specific in their clinical applications.9 ,10
The ‘ideal biomarker’ of neonatal infection and necrotising enterocolitis (NEC)
A. Clinical properties
1. The ideal biomarker should possess the following clinical characteristics:
a well-defined optimal cut-off level
sensitivity and negative predictive value approaching 1.00, and specificity and positive predictive value >0.85.
Notes: In contrast, a biomarker with high specificity and positive predictive value is good for ‘ruling in’ infection/NEC
2. The ideal biomarker (or a panel of biomarkers) should provide an algorithm or a probability score (±taking into account clinical signs and symptoms) for guiding neonatologists whether to initiate antibiotic treatment (eg, ApoSAA score).
3. Detect infection at a very early stage
daily surveillance and identify the condition before clinical manifestations (eg, neutrophil CD64)
OR
diagnostic at clinical presentation (‘early-warning’ biomarkers, eg, interleukin (IL)-6, interferon-gamma inducible protein 10 (IP-10), neutrophil CD64, etc).
4. Predict the severity of infection at the onset of clinical presentation, (eg, using IL-10, IL-6 and regulated upon activation normal T cell expressed and secreted (RANTES) sequentially).
5. Differentiate different categories of pathogens: (i) virus versus bacteria versus fungus, or (ii) Gram-positive organisms versus Gram-negative organisms (eg, Gram-specific qPCR test), within the first few hours of clinical presentation.
6. Diagnose a specific disease entity or identify specific organ injury, such as NEC and bowel injury (eg, ‘enhanced non-specific’ biomarkers, eg, faecal calprotectin and S100A12, and gut-associated biomarkers, eg, L-FABP, I-FABP, TFF-3 and LIT score).
7. Monitor disease progress, guide antimicrobial treatment and detect development of complications, such as formation of intra-abdominal abscesses or an inflammatory mass of matted necrotic bowel (eg, serial C-reactive protein (CRP) measurements).
8. Predict prognosis and mortality (eg, combined use of IL-6 and IL-10).
B. Laboratory properties
1. A sustained increase or decrease in level of a stable biomarker allows an adequate time window (12–24 h) for specimen collection, storage and laboratory processing before decomposition of the active mediator (eg, neutrophil CD64, CRP).
Notes: Unstable biomarkers with very short half-life are not clinically useful as neonatologists may not catch the peak or trough level during specimen collection
2. Very small volume of specimen (eg, 0.05 mL whole blood for neutrophil CD64 or 0.05 mL plasma for a panel of >10 chemokines/cytokines measurement).
3. Excellent agreement of biomarker levels between capillary and venous samples, especially for daily surveillance (eg, neutrophil CD64).
4. Automatic laboratory measurement.
5. Quantitative measurement of biomarker level.
6. Quick turnaround time from specimen collection to reporting of results (eg, <6 h in Gram-specific qPCR test or neutrophil CD64 measurement).
7. Daily or on-demand availability of the test in routine clinical laboratories.
8. Low cost.
I-FABP, intestinal-fatty acid binding protein; L-FABP, liver-fatty acid binding protein; TFF-3, trefoil factor-3.
Classes of laboratory biomarkers
The characteristics of non-specific mediators of inflammation have been described in detail in our previous studies, and they are broadly divided into three main categories:1 (i) acute-phase proteins, for example, C-reactive protein (CRP) and serum amyloid A (SAA) or its variants; (ii) chemokines and cytokines, for example, regulated upon activation normal T cell expressed and secreted (RANTES), interleukin (IL)-6, IL-8, IL-10; and (iii) cell surface antigens, for example, neutrophil CD64 and CD11b. Although the majority of these biomarkers are components of the inflammatory cascade, different classes of mediators have varied clinical and laboratory characteristics.1 In general, these mediators will respond to any event that may trigger an inflammatory host reaction, including systemic infection, NEC, surgery, trauma and chemical injury to tissues and organs, for example, meconium aspiration syndrome.1 Many of these aetiologies can be easily identifiable from the clinical history. Furthermore, with the absence of occult non-infective causes of inflammation such as inflammatory bowel diseases and connective tissue diseases in this age group, an upregulation or downregulation of their circulating levels would most likely be due to systemic infection or NEC. The major drawback of these non-specific inflammatory biomarkers would be their inability to identify local or compartmentalised infections,1 particularly urinary tract infection, mild pneumonia and, on rare occasions, even early fungal meningitis.
The ‘omic’ approach
With recent advances in medical technologies, investigators are using multi-omic approaches such as metabolomics,11 metagenomics12 and proteomic13 for comprehensive and systematic detection of differentially expressed mediators in the form of unique molecular or protein signatures for revealing mechanisms of diseases, host versus pathogen interaction, and microbiota/microbial ecology.12 ,14 These approaches could also be targeted for discovery of novel biomarkers13 so as to diagnose, monitor and stratify tailor-made treatment for a specific condition and thus, paving the way for personalised neonatal medicine in future. The formulation of the ApoSAA score from SAA and apolipoprotein-CII is a prime example of using the proteomic technique to discover a panel of novel biomarkers for early and accurate diagnosis of neonatal sepsis/NEC.13 Nonetheless, most omic studies in neonatology are either in the pioneering phase or proof-of-concept experimental stage. It is, however, envisaged that with new emerging technologies and advanced bioinformatics techniques, novel and specific biomarkers will be identified in the coming decades for important diseases, both in neonates and adults.
Clinical use of laboratory biomarkers
Biomarker discovery and use have witnessed advancements from simple diagnosis of systemic infection to more sophisticated objectives.
Early and accurate diagnosis of systemic infection/NEC
Early and accurate identification of neonatal infection/NEC are the prime objectives of infection biomarkers.1 The ApoSAA score could guide the initiation of antimicrobial treatment in suspected cases of neonatal sepsis/NEC. By stratifying infants into different risk categories, it was possible to withhold antibiotics in 45% of infants with suspected sepsis/NEC and allowed discontinuation of treatment within 24 h in 16% of non-sepsis infants.13 The negative predictive value of this antibiotic screening policy was 100% and the score would not miss a single case of infection/NEC that had not been commenced on antimicrobial treatment.13 Another commonly used strategy is to combine an ‘early-warning’ biomarker (eg, neutrophil CD64 or ILs) with a ‘late-warning’ biomarker (eg, CRP) for diagnosing infection/NEC at clinical presentation, as neonatologists have no knowledge on what stage of the infection/NEC is correlated with the initial blood sampling.1 ,2 ,15 Thereafter, serial measurements with a ‘late-warning’ biomarker or an all-round biomarker (eg, neutrophil CD64) at 24 and 48 h after the initial presentation would assist accurate confirmation of infection/NEC and allow stoppage of antibiotics in non-sepsis/non-NEC cases. A persistent elevation of circulating CRP level can indicate inadequate treatment due to microbial antibiotic resistance or emergence of complications, for example, development of intra-abdominal abscess or formation of an inflammatory mass secondary to fusion of matted necrotic bowel in NEC.6 However, localised infections such as urinary tract infection, mild pneumonia or loculated abscess may be missed by these non-specific systemic mediators,1 and therefore exercising sound clinical judgement is vital for effective interpretation of biomarker results.
Daily surveillance
It is now recognised that the intense proinflammatory host response is a major component of neonatal sepsis/NEC and such immunological reactions can adversely affect distant organs, especially the central nervous system. By identifying infection/NEC before clinical manifestations, neonatologists hope to treat the underlying condition early and maximise the chance for prompt initiation of antibiotics that may potentially dampen the exaggerated immunological response at a very early stage. Daily surveillance with an ‘early-warning’ biomarker offers the best opportunity for detecting infection/NEC before clinical manifestations. As surveillance biomarkers should possess additional laboratory properties, including very small blood volume requirement permitting daily assay, short laboratory turnaround time and good agreement of results between capillary and venous samples,3 such mediators are difficult to find and surveillance studies for neonatal infection/NEC are notoriously difficult to perform. Thus far, only a few trials have purposefully evaluated various categories of biomarkers as surveillance indicators. In a prospective surveillance study involving 101 very low birth weight (VLBW) infants, IL-6 and IL-1 receptor antagonist (IL-1ra) were found to be significantly upregulated as early as up to 2 days before clinical diagnosis of sepsis.4 Sensitivities on day −1 (ie, 24 h before clinical presentation) for IL-6 and IL-1ra were 0.57 and 0.64, respectively.4 The investigators suggested that the use of IL-6 and IL-1ra should permit earlier commencement of antibiotics that could result in improved outcomes. In another study involving 30 extreme LBW infants with daily blood sampling, CD11b expression gradually increased over the 3 days prior to sepsis evaluation and before arousal of clinical suspicion of sepsis.5 Our research team has recently performed a daily surveillance study involving 146 VLBW infants.3 There was excellent agreement of neutrophil CD64 levels between capillary and venous samples (r=0.999).3 Neutrophil CD64 was able to detect systemic infection/NEC 1.5 days before clinical presentation (sensitivity, specificity, positive and negative predictive values were 0.89, 0.98, 0.99 and 0.41, respectively).3 Unexpectedly, a group of infants with asymptomatic CD64 activation was identified. These asymptomatic cases followed a similar pattern of activation resembling genuine infection/NEC but with a more gradual rise and fall in CD64 levels that occurred over a shorter 3-day period and with much lower peak levels.5 All infants recovered spontaneously without antibiotic treatment. We speculated that these episodes represented subclinical infection secondary to transient bacterial translocation or mild viral infection rather than spurious laboratory measurements.5 Despite the favourable sensitivity and specificity, the use of neutrophil CD64 as a surveillance biomarker would translate into performing an additional 41% of unnecessary sepsis workups for exclusion of ‘false-positive’ cases due to asymptomatic CD64 activation.3 Faecal calprotectin and S100A12 have also been studied as screening biomarkers of NEC. The large variation in stool concentrations between individual patients rendered these proteins insensitive as screening tools.16 ,17 To date, the laboratory criteria for surveillance biomarkers have been well defined (box 1), but investigators have yet to find an ideal biomarker for screening purpose. Also, it would be important to take into account of asymptomatic activation cases3 that were not reported in the first two studies.4 ,5
Bedside monitoring of physiological parameters, such as electrocardiographical characteristics for identifying atypical patterns of decreased heart rate variability and/or transient deceleration, could assist in early detection of sepsis/NEC.18–20 Such data could further be transformed into an index or scoring system that would be useful for quantifying the frequency and magnitude of abnormalities.20 In future, the use of physiological data coupled with surveillance biomarkers could improve the diagnostic utilities in detecting neonatal sepsis/NEC. Non-invasive faecal or urinary biomarkers are probably more suited for daily surveillance purposes. Ultimately, algorithms using both clinical and laboratory data could be developed in the form of an ‘app’ for mobile electronic devices to guide frontline neonatologists at the bedside on the ‘probability’ or ‘likelihood’ of a particular infant developing sepsis/NEC.
Rapid bedside point-of-care test
The development of a bedside point-of-care test for rapid identification of neonatal sepsis has always been one of the objectives in biomarker research. The concept has already been proven to be feasible using a miniaturised protein microarray chip that could simultaneously quantify nine cytokines/chemokines and acute-phase proteins, including IL-6, IL-8, IL-10, tumour necrosis factor-α (TNF-α), S-100, procalcitonin, E-selectin, CRP and neopterin, with only 4 µL of serum.21 The measurement could be completed within 2.5 h using a single-step assay. Recently, an IL-6 bedside point-of-care test with results available within 1 h has been field tested for quick detection and exclusion of neonatal bacterial sepsis.22 The sensitivity of the bedside IL-6 test was apparently comparable with that of the laboratory IL-6 measurement for detecting late-onset sepsis, but less sensitive for identifying early-onset cases.22 The ultimate aim would be for neonatologists to collaborate with industrial partners to develop portable point-of-care machines that could use minimal volume of biofluid to accurately measure ‘early-warning’ biomarkers for bedside identification or exclusion of bacterial sepsis/NEC.
Predicting the severity of infection/NEC and mortality
In general, the greater the magnitude of change in inflammatory mediator concentrations from baseline levels, the more severe is the infection, bowel inflammation or virulence of the causative pathogen. However, very few studies have purposefully attempted to evaluate biomarkers that could objectively reflect the severity and mortality of neonatal sepsis/NEC.7 ,8 As the proinflammatory and anti-inflammatory counter-regulatory mechanism is likely to be operational in VLBW infants,8 high circulating IL-6, IL-10 and TNF-α concentrations, and IL-10:TNF-α ratio were reported in cases with severe sepsis. Although a transient increase in mediator levels did not necessarily infer a poor prognosis,8 the IL-6:IL-10 ratio was disproportionally increased in deceased infants at the onset of clinical presentation and continued to rise despite appropriate treatment. This phenomenon confirmed the suspicion that a delicate balance existed between the two regulatory pathways and the ratio could reflect the severity of the condition as well as enabling neonatologists to predict the outcome. In addition, the sequential use of IL-10, IL-6 and RANTES in a mathematic model could predict the subsequent development of disseminated intravascular coagulation in VLBW infants with severe sepsis/NEC at the initial stage of clinical presentation.7 The sensitivity, specificity, positive and negative predictive values of this panel of biomarkers were 1.00, 0.97, 0.85 and 1.00, respectively. Thus, these mediators are able to assist neonatologists in identifying the sickest infants who are most in need of urgent treatment and those who require escalation of treatment at the onset of sepsis, so that critically ill patients who are most at risk of dying could be specifically targeted.7 In the era of personalised medicine, the inclusion of prognostic biomarkers is likely to become a prerequisite component in the management of intensive care infants.
Identification of specific disease, organ injury or pathogen
Non-specific mediators of the inflammatory cascade are unable to differentiate sepsis from NEC, or sepsis caused by different types of organisms. This differentiation is, however, important in the daily clinical setting as different causative pathogens and diseases are associated with very different management strategies (eg, choice of antibiotics, duration of parenteral nutrition and the need for surgical intervention) and outcomes. In the past decade, many investigators have focused on searching for specific biomarkers that could pinpoint a specific organism (or a category of organisms) and disease (or specific organ injury).
Laboratory biomarkers of NEC could be classified into two main functional categories, namely ‘enhanced non-specific’ biomarkers, for example, faecal inflammatory mediators, and gut-associated biomarkers, for example, gut-derived proteins. ‘Enhanced non-specific’ biomarkers of NEC are inflammatory mediators that are involved in the common inflammatory cascade, but by virtue of the nature of the specimen (ie, stool) localises the inflammatory process to the GI tract. Calprotectin (S100A8/S100A9 heterodimer) and S100A12 are proteins secreted in abundance by neutrophils via the inflammatory pathways. Both are resilient to bacterial degradation. However, the wide variation of calprotectin concentrations in faecal matters (9–1867 µg/g of faeces),16 dependence on gestational and postnatal age,23 widely variable cut-off levels (200–2000 µg/g of faeces) for diagnosis of NEC among studies10 ,24 ,25 and paradoxical decrease in concentration (<24 µg/g of faeces) in fulminant cases,23 have severely undermined its usefulness as a diagnostic or surveillance biomarker of NEC. Likewise, there is much overlap in faecal S100A12 concentrations between NEC and non-NEC infants.17 Other disadvantages of using faecal specimens include non-homogeneous dispersion of the mediator in a semi-solid medium, the unequal distribution being further aggravated by the focal and segmental nature of gut inflammation in NEC, and difficulty in obtaining a representative stool specimen with intestinal ileus. More studies are warranted to further delineate the role of inflammatory proteins and other faecal compounds, for example, volatile organic compounds for surveillance and early detection of mucosal bowel injury and/or NEC in infants.
Gut-associated biomarkers are peptides that are predominantly produced by intestinal cells or are part of the constitutional components lining the lumen of the GI tract. Inflammation and injury to the intestinal mucosa could result in liberation of large quantities of such proteins into the bloodstream that are then excreted in urine.9 ,10 ,26–28 In a case–control study, circulating liver-fatty acid binding protein (L-FABP), intestinal (I)-FABP and trefoil factor-3 (TFF-3) as individual biomarkers or in combination as the LIT score have been identified as specific biomarkers of NEC.9 In particular, the LIT score was useful in differentiating severe cases of surgical NEC from those with only septicaemia or disease-free control infants. The sensitivity and specificity were 0.83 and 1.00.9 Also, the LIT score was useful in differentiating severe surgical NEC from milder non-surgical cases, as well as non-survivors from survivors.9 However, these gut-associated biomarkers were not effective in identifying mild NEC cases and hence, would not be useful as surveillance or ‘early-warning’ biomarkers.9 Similarly, urinary gut-associated proteins:creatinine ratio was useful for detecting non-survivors and surgical cases.10 ,26 Unfortunately, urine samples could be difficult to collect in female preterm infants and critically ill patients with concomitant renal failure. The short plasma half-life of these proteins implies that sustained and severe damage of the bowel is required to maintain a significantly elevated plasma or urinary level and thus, explains why only infants with severe NEC are correctly identified. Also, there are continuous turnover and liberation of gut-associated proteins into the bloodstream from the normal bowel. We postulate that in mild focal segmental NEC, the relatively small quantity of protein released into the circulation could be masked by their pre-existing background concentrations, thereby resulting in non-appreciable increase in plasma/urinary levels. Overall, these biomarkers serve as important indicators for assessing the severity of NEC and are useful for strategic planning of surgery during the early phase. Recently, targeted liquid chromatography/mass spectrometry approach has identified a panel of seven proteins (α2-macroglobin-like protein 1, cluster differentiation protein 14, cystatin 3, fibrinogen α chain, pigment epithelium-derived factor, retinol binding protein 4 and vasolin) in urine suitable for differentiating NEC from sepsis and surgical from non-surgical NEC.29 However, further validation of this signature for clinical usefulness is required.
In the past decade, neonatologists have attempted to differentiate between Gram-negative and Gram-positive bloodstream infection before culture results are available. Early and correct identification of the causative Gram-specific bacteria and appropriate use of antibiotics are vital for successful treatment of septicaemia. In a recent study involving 218 suspected infection episodes, the Gram-specific probe-based quantitative PCR (qPCR) test has sensitivity and specificity of 0.86 and 0.99 for identifying Gram-negative and 0.74 and 0.99 for Gram-positive septicaemia, respectively.30 The results have been supported by similar studies.31 More importantly, despite negative blood cultures in infants with intra-abdominal sepsis, for example, peritonitis, the qPCR test could identify the Gram-specific category of causative pathogens in the peritoneal fluid and blood.30 The qPCR test has been more useful for detecting Gram-negative than Gram-positive organisms due to difficulty in breaking up the cell wall for DNA extraction in the latter group of bacteria.29 The high specificity of the qPCR test renders it an excellent tool for ‘ruling in’ neonatal infection.30 Recently, a new molecular multiplex PCR system has also been demonstrated to have high sensitivity (0.91) and specificity (0.80) for rapid detection of nosocomial sepsis in preterm infants and could therefore serve as an useful adjunct diagnostic test in addition to the gold standard of blood culture.32
Summary
In contrast to previous reviews of infection biomarkers that focus mainly on different categories of mediators with varied biochemical characteristics,1 this article primarily concerns the clinical usage of biomarkers for daily surveillance, bedside point-of-care testing, diagnosis and prediction of the severity and prognosis of infants with sepsis/NEC. It also highlights the directions of future research in biomarker studies. With recent advances in ‘multi-omic’ technologies, more novel biomarkers will certainly be discovered in the future. In the era of personalised medicine and targeted therapy for individual diseases, the emergence of specific mediators for diagnosis of specific diseases and to gauge the magnitude of organ injury will undoubtedly be the next vital step for embracing the development of biomarkers in neonatology.
References
Footnotes
Contributors PCN: principal author of the first draft, designed the layout and decided on the appropriate contents to be included in this review. TPYM: assisted in literature search, constructed the table and amendment of the article. HSL: assisted in literature search, modifying materials to be included in the review, amendment and proofread the article.
Funding None.
Competing interests None.
Provenance and peer review Commissioned; externally peer reviewed.