Biochemical markers of neonatal sepsis

Pathology (February 2008) 40(2), pp. 141–148

NEONATAL AND PAEDIATRIC SEPSIS

Biochemical markers of neonatal sepsis HUGH S. LAM

AND

PAK C. NG

Department of Paediatrics, Prince of Wales Hospital, The Chinese University of Hong Kong

Summary The use of biochemical markers in neonatal infection has remained an important area of research in the past decades. Many infection markers are components of the inflammatory cascade and reflect the host’s immunological status and response to infection. Cytokines and chemokines such as interleukin (IL)-6 and IL-8 have been demonstrated to have good diagnostic utilities as early phase markers, while acute phase reactants such as C-reactive protein and procalcitonin have superior diagnostic properties during the later phases. Other markers, including inter-a-inhibitor proteins, IL-10 and regulated upon activation normal T cells expressed and secreted (RANTES) have been demonstrated to yield important prognostic information and may help the clinician identify infants who will develop fulminant infection from the outset of presentation. The advent of flow cytometry and molecular techniques have made crucial contributions to the field and promise to further improve the diagnostic accuracy and clinical management of infected infants. Key words: Chemokines, cytokines, infection markers, leukocyte surface antigens, neonatal infection. Abbreviations: CRP, C-reactive protein; GRO, growth-related oncogene; IFN, interferon; IL, interleukin; LBP, lipopolysaccharide-binding protein; PCT, procalcitonin; RANTES, regulated upon activation normal T cells expressed and secreted; SAA, serum amyloid A; TGF, transforming growth factor; TNF, tumour necrosis factor; VLBW, very low birth weight. Received 12 August, revised 7 October, accepted 8 October 2007

INTRODUCTION Investigation of biochemical markers for neonatal infection is an important area of research, because morbidity and mortality in neonatal sepsis remains substantial1,2 despite continued advances in neonatology3 and choices of novel antibiotics. Most current biochemical markers are derived from components of the intricate and complex inflammatory response to invading pathogens. This review focuses on recent studies of chemokines, cytokines, cell surface antigens and acute phase proteins as potential infection markers. The role of other modalities of investigation of neonatal infection will also be addressed. Table 1 contains a glossary of infection markers discussed in this review.

ROLE OF INFLAMMATORY MEDIATORS Inflammatory mediators play important roles in the host response to pathogens. These molecules can have pro-

inflammatory or anti-inflammatory properties. Proinflammatory mediators, such as interleukin (IL)-1b, IL-6 and tumour necrosis factor-a (TNF-a) activate host defences against infective agents, while anti-inflammatory mediators, such as IL-4, IL-10 and transforming growth factor b (TGF-b), are important in regulating and limiting the inflammatory response, preventing an excessive reaction which may itself cause host organ damage and tissue cell death.4–6 Over the past decades, research into various actions and interactions of these mediators has further increased our understanding of the inflammatory cascade and provided us with new avenues of investigation into their clinical application. One such important area is the field of neonatal infection. Despite advances in neonatal intensive care,3 stringent infection control measures, and an increasingly wide range of newer generations of antimicrobial agents, neonatal infection remains an important cause of morbidity and mortality amongst neonates, especially very low birth weight (VLBW) and extremely premature infants.7,8 Early identification of infected neonates is fraught with difficulties as the clinical features during the early phases of infection may be non-specific and inconspicuous. Routine microbiological and haematological investigations that are currently in common use have substantial limitations, e.g., the wide normal variation of total and differential white cell counts, subjective nature of white cell morphology, difficulty in obtaining culture specimens from various tissues, low sensitivity of blood culture in neonates and often long periods of time required for clinically useful results by conventional culture techniques. The development of newer biochemical markers will further increase the ability of neonatal clinicians to differentiate between infected and non-infected infants. A comprehensive account of all the biochemical markers that have been studied in the past decades is beyond the scope of this review and, thus, only a select collection of the more important and promising recent infection markers will be described.

APPLICATIONS OF BIOCHEMICAL MARKERS Each of the plethora of infection markers, e.g., acute phase reactants, chemokines, cytokines and leukocyte cell surface antigens, have distinct characteristics. The distinct properties of various biochemical markers suggest that each mediator has a different set of indications and restrictions when applied to different types of infection (e.g., viral versus bacterial), or even different phases of the infective process.9 Infection markers have traditionally been used to assist neonatologists decide on which patients to

Print ISSN 0031-3025/Online ISSN 1465-3931 # 2008 Royal College of Pathologists of Australasia DOI: 10.1080/00313020701813735

142 TABLE 1

LAM AND NG

Pathology (2008), 40(2), February

Glossary of potential diagnostic markers in neonatal sepsis

Pro-inflammatory cytokines IFN-g Interferon-g: predominantly produced by Th1 cells and facilitates activation of pro-inflammatory cells, e.g., macrophages IL-1b Interleukin-1b: mainly produced by macrophages to help regulate the pro-inflammatory process IL-6 Interleukin-6: predominantly produced by leukocytes and hepatocytes in response to infection and trauma TNF-a Tumour necrosis factor-a: mainly produced by macrophages in response to bacterial and other inflammatory products Anti-inflammatory cytokines IL-4 Interleukin-4: predominantly produced by Th2 cells IL-10 Interleukin-10: inhibits production of pro-inflammatory cytokines by Th1 cells TGF-b Transforming growth factor b: regulates cellular proliferation and differentiation, and suppresses T helper cell function CC chemokines MCP-1 RANTES CXC chemokines GRO-a IL-8 IP-10 MIG

Monocyte chemoattractant protein-1: attracts natural killer cells and activates mast cells Regulated upon activation normal T cells expressed and secreted: attracts eosinophils, lymphocytes and monocytes Growth-related oncogene-a: attracts neutrophils and suppresses myeloid colony formation Interleukin-8: attracts neutrophils and stimulates phagocytic activity Interferon-g-inducible protein-10: attracts activated T lymphocytes, and may be induced by interferon-g. It has anti-tumour activity and a regulatory role in angiogenesis Monokine induced by interferon-g: biological activity similar to IP-10

Acute phase reactants CRP C-reactive protein: an acute phase protein produced by the liver that increases in response to IL-6 IaIp Inter-a-inhibitor proteins: a group of plasma proteins that inhibit serine proteases and possess anti-inflammatory properties LBP Lipopolysaccharide-binding protein: an acute phase protein that is mainly produced by the liver PCT Procalcitonin: the precursor of the peptide hormone calcitonin mainly produced by the liver and by monocytes SAA Serum amyloid A: a group of structurally related proteins that are released in response to injury and infection by a broad range of cell types, e.g., hepatocytes, smooth muscle cells, endothelial cells, monocytes Leukocyte surface antigens CD11b Cluster of differentiation 11b: a b2-integrin adhesion molecule present on leukocyte cell surfaces and binds molecules such as complement components and lipopolysaccharide CD64 Cluster of differentiation 64: a receptor found on leukocyte cell surfaces that binds the Fc portion of IgG antibodies with high affinity

discontinue antibiotics early.9,10 In recent years, however, there has been new insight into how these markers should be best utilised, especially in infants who appear to be clinically stable,11,12 or whether to start antibiotic treatment when daily monitoring of markers indicate an abnormal level for seemingly healthy infants.13 Other applications of infection markers that have been studied recently include not only their utilities as diagnostic tools but also their provision of vital prognostic information.4,14,15 Cytokines and chemokines Inflammatory cytokines and chemokines, such as IL-1b, IL-6, IL-10, interferon g (IFN-g), TNF-a, IL-8, regulated upon activation normal T cells expressed and secreted (RANTES), monokine induced by interferon-g (MIG), monocyte chemoattractant protein-1 (MCP-1), growthrelated oncogene-a (GRO-a) and interferon-g-inducible protein-10 (IP-10), mediate the host response to infection in complex inflammatory pathways. Varying levels and ratios of these mediators, therefore, provide the clinician with valuable diagnostic and prognostic information regarding the status of the inflammatory process.4,6,15,16 Cytokines Of the pro-inflammatory cytokines, IL-6 has been one of the most widely studied for its potential as an infection marker in neonatal infection.4,17–24 It is produced by both T and B cells.25 It serves many functions, including regulation of the host response to infection.26 Exposure of the host to bacterial products results in a rapid and substantial increase in blood IL-6 concentrations.16,20,27 IL-6 in turn stimulates hepatocytes to produce acute phase reactants such as C-reactive protein (CRP).25 Therefore, IL-6 is potentially a more useful marker than CRP during

the early phase of infection with a sensitivity of 89 versus 60% at the onset of clinical suspicion of nosocomial neonatal infection, respectively.19 More importantly, in the same study, the negative predictive value of IL-6 (91%) was much higher than CRP (75%). This early rise in blood concentrations in response to infection has also been demonstrated in early-onset neonatal infection. In a study where cord blood IL-6 levels were measured, it was found that sensitivities and negative predictive values remained high (87–100% and 93–100%, respectively).28,29 In view of the short half-life of IL-6, successful treatment of infection leads to rapid reduction in its circulating concentrations to undetectable levels by 24 hours.6,16,20 Although this characteristic suggests that IL-6 can be used to monitor the host response to treatment and progress of infection, it also implies that the window of opportunity for specimen sampling is narrow. Whilst the sensitivity of CRP increases to 82% by 24 hours and 84% by 48 hours after the onset of infection, the corresponding parameters for IL-6 decrease to 67% and 58%, respectively.19 In view of these limitations, combining IL-6 with markers that are more sensitive during the later phases of infection has been studied to improve the diagnostic utilities.19,24 For example, the combination of IL-6 and TNF-a is more sensitive than IL-6 or TNF-a alone in early-onset neonatal infection (98.5, 90 and 87.9%, respectively),24 while various combinations of IL-6, CRP and TNF-a can maintain sensitivities and negative predictive values above 90% for late-onset infection from presentation until 48 hours afterwards.19 The inflammatory process is regulated not only by proinflammatory chemokines and cytokines, but also antiinflammatory mediators. Anti-inflammatory cytokines, such as IL-10 and transforming growth factor b (TGF-b)

BIOCHEMICAL MARKERS OF NEONATAL SEPSIS

contribute to the compensatory anti-inflammatory response syndrome, which helps prevent an exaggerated proinflammatory response.30 Compared with adults, the capacity of both preterm and term neonates to produce inflammatory cytokines, as measured by IL-10 mRNA expression, IL-10 production and TGF-b-positive lymphocytes, has been shown to be significantly reduced.6,30 It has been suggested that the anti-inflammatory response in the neonatal population is immature, thereby increasing susceptibility to organ damaging effects of excessive systemic inflammatory response syndrome.6 Therefore, anti-inflammatory cytokines have been investigated in recent studies with regard to their prognostic capability.4,5,31 Raised IL-10 to TNF-a ratios in adults with community acquired febrile disease were associated with a worse prognosis.5 In neonates, the situation is less certain. In a small study involving VLBW infants, raised IL-10 to TNF-a ratios were associated with severe infection, but not necessarily poorer prognosis.4 In the same study, the infant who died exhibited an increased IL-6 to IL-10 ratio from the outset of infection. Whether neonates with an imbalanced pro-inflammatory/antiinflammatory cytokine response at the outset of infection have a worse prognosis remains to be confirmed by a larger population. However, it appears that an intense proinflammatory and anti-inflammatory response represents severe infection and can predict the development of disseminated intravascular coagulopathy at initial presentation of illness.18 Chemokines Of the many studies on chemokines, IL-8 is one of the most extensively investigated in neonatal infection.11,12,32–35 IL-8 is the only interleukin which belongs to the chemokine families, a group of chemoattractant cytokines which regulate leukocyte migration and activation.19,36,37 Similar to IL-6, IL-8 rises promptly within 1–3 hours of infection and has a short half-life of less than 4 hours.35 Many studies have investigated the use of IL-8 as an early-phase infection marker.35,36 Fotopoulos et al. found that IL-8 levels were significantly higher in infants diagnosed with perinatal infection than in controls on day 1 (median serum concentrations were 348 pg/mL and 0 pg/mL, respectively), but not on day 4.36 In another study, the sensitivity of IL-8 was much higher than CRP within 6 hours of suspected early-onset bacterial infection (71 versus 14%, respectively),35 suggesting that it is superior to CRP in early phases of infection. The diagnostic utilities of IL-8 can be further enhanced by appropriate processing of the blood sample.33,34,35 As IL-8 is rapidly bound by leukocyte receptors, plasma IL-8 reflects only a small proportion of total IL-8 secreted in response to infection. It has been shown that by processing whole blood with detergent, cell-associated IL-8 can be released and the total IL-8 pool measured.33,34 After this special processing technique, it was found that at 6 hours after clinical suspicion of infection, the sensitivity and negative predictive values were increased from 71 to 97% and 89 to 99%, respectively.35 The values at 24 hours were also increased from 33 to 70% and 66 to 80%. Another important implication of such findings is that the window of opportunity for using IL-8 as an infection marker is increased.

143

Traditionally, neonatologists have used infection markers to decide when to discontinue antibiotics early.10 More recently, investigators have studied the possibility of using IL-8 in combination with other infection markers as part of a decision algorithm to assist neonatal clinicians determine whether to commence antibiotic therapy in specific groups of newborns suspected of suffering from bacterial infection.11,12 Franz et al. utilised IL-8 and CRP as part of a diagnostic algorithm in a multicentre randomised controlled trial.11 Neonates suspected to suffer from mild to moderate early-onset bacterial infection were recruited. Infants randomised into the standard group received treatment according to the centre’s usual protocols, whereas infants in the IL-8 group received antibiotics if either IL-8 or CRP concentrations exceeded the predetermined values. While the proportion of infants treated in the IL-8 group (36.1%) was significantly less than the standard group (49.6%), the proportions of infected infants missed was not significantly different (14.5 and 17.3% in the IL-8 group and standard group, respectively). The findings of this study represent a crucial step forward in the strategy for using biochemical markers in the area of neonatal infection. Apart from IL-8, the roles of other chemokines, such as IP-10, MIG, MCP-1, RANTES and GRO-a as infection markers in early and late-onset neonatal infection have been studied.36,38 Most but not all chemokines belong to either the CC or CXC groups depending on the structure of the terminal cysteine residues. CXC chemokines differ from CC chemokines by the separation of the two amino terminal cysteines, ‘C’, by an amino acid, designated ‘X’. IL-8, IP-10 and MIG are CXC chemokines responsible principally for recruitment of neutrophils and activated T lymphocytes, whilst RANTES and MCP-1 are CC chemokines responsible mainly for attracting monocytes and T lymphocytes.36 A study by Fotopoulos et al. showed that IP-10 was raised in infants on day 1 suffering from either perinatal asphyxia (median 81 pg/mL; interquartile range 42.5–181.3 pg/mL) or perinatally acquired infection (232.7; 144.7–500.0 pg/mL) compared with controls (60.7; 32.4–84.5 pg/mL). In contrast, RANTES was significantly decreased in perinatally acquired infection (32 181; 22 612–41 910 pg/mL) versus controls (50 280; 43 765–69 415 pg/mL), but not significantly different in infants suffering from birth asphyxia. Such difference between infected and non-infected infants persisted until day 4.36 In a recent study investigating a panel of chemokines for use in late-onset neonatal infection, IP10 was found to show high sensitivity, specificity and negative predictive value at the outset of infection presentation (93, 89 and 97%, respectively), but with a slight decrease in diagnostic utilities 24 hours thereafter (88, 83 and 95%, respectively).38 The use of chemokines in combination with other inflammatory cytokines as prognostic indicators have also been investigated.18 The sequential measurement of IL-10, IL-6 and RANTES at the onset of clinical suspicion of infection could predict disseminated intravascular coagulation with high sensitivity, specificity, and negative and positive predictive values (100, 97, 100 and 85%, respectively). While the prognostic capability or diagnostic utilities of certain chemokines such as IP-10 and RANTES in neonatal infection seem promising, further confirmation into how these mediators

144

LAM AND NG

can facilitate the clinical management of infected neonates is warranted. Acute phase reactants Another group of biochemical markers that have been widely investigated are acute phase reactants.19,22,23,39–41 The diagnostic utilities of acute phase proteins and their combinations at early and late phases of infection for various types of infection have been studied. C-reactive protein CRP serves as a good example of how such biochemical markers aid the neonatal clincian in distinguishing infected from non-infected infants. In view of its late up-regulation, taking approximately 8–10 hours for synthesis,42 it is not useful as an early phase infection marker but has good diagnostic utilities in the later phase as both sensitivity and negative predictive value increase from 0 to 24 hours after onset of suspected infection (14–100% and 75–100%, respectively).35 Procalcitonin Procalcitonin (PCT) is the peptide prohormone of calcitonin, produced predominantly by monocytes and hepatocytes. Its circulating concentration increases substantially within 2 hours of infection and is one of the most extensively investigated acute phase proteins.40 This early rise in circulating concentrations renders it more useful as a diagnostic marker than CRP in early phases of the infection. However, its use in early-onset infection is hampered, as procalcitonin is often physiologically elevated during the first 2 days of life.39,43 Further, non-infective perinatal events, such as intracranial haemorrhage, perinatal asphyxia and maternal pre-eclampsia, may also increase circulating PCT concentrations.23,40 Thus, the perinatal history must be taken into account in conjunction with a nomogram39 before PCT concentrations can be properly interpreted during the first few days of life. An earlier study in neonates demonstrated excellent diagnostic utilities for PCT during early-onset infection (sensitivity 92.6%, specificity 97.5%, positive predictive value 94.3%, negative predictive value 96.8%), and even better utilities in late-onset infection (sensitivity and specificity both 100%).43 Despite variations in the diagnostic utilities of PCT in the literature, a study comparing PCT against IL-6 and CRP still concluded that the sensitivity of PCT for early-onset infection during the first 12 hours of life was best (PCT 77%, IL-6 54% and CRP 69%), while the specificity remained favourable (91, 100, and 96%, respectively).22 Similarly, the diagnostic utilities of PCT in late-onset infection were also better than many other markers: sensitivity and specificity were 69 and 89% for PCT, 65 and 52% for CRP, 68 and 76% for IL-6, and 84 and 52% for IL-8.16 Further, the diagnostic utilities of PCT, in contrast to IL-6, seem to be unaffected by the severity of illness and risk status of the infant, rendering it a more robust biochemical marker of infection than IL-6.23 It has been shown by Verboon-Maciolek et al. that circulating PCT concentrations increased rapidly at the onset of infection, then decreased and normalised within 2–3 days of effective treatment.16 This is in contrast to other early phase infection markers such as IL-6 and IL-8, which became undetectable within 24 hours of appropriate treatment,16,20,35 and late phase markers such as CRP,

Pathology (2008), 40(2), February

which took 5–6 days to normalise after coagulase-negative Staphylococcus infection.16 In the same study, it was shown that PCT was more specific to bacterial infection than CRP; while CRP levels were elevated in 40% of infants suffering from enteroviral infection, PCT levels were only increased in 20%. Some investigators have attempted to avoid the confounding effects of perinatal events on PCT concentrations by measuring umbilical cord blood.44 Cord blood PCT was found to be significantly more sensitive and specific (87.5 and 98.7%) than cord blood CRP (50 and 97%, respectively) in detecting perinatal infection in newborns. The same study also suggested that cord blood PCT could help the neonatologist distinguish between infants who were genuinely infected from those with only bacterial colonisation. Serum amyloid A Serum amyloid A (SAA) is another promising acute phase reactant that has been investigated recently. It has a diversity of sources, including hepatocytes, smooth muscle cells, endothelial cells and monocytes, and is released in response to injury and infection. It has many roles in the inflammatory process and can stimulate the secretion of IL-8 from neutrophils.45 Circulating SAA concentrations have been shown to increase with age, with lowest levels in umbilical cord blood and highest in the serum of elderly patients.46 The interpretation of SAA levels must, therefore, be adjusted for age of the subject. There are promising results from a recent study of lateonset neonatal infection by Arnon et al.41 In this study, SAA was compared against other biochemical markers, including IL-6 and CRP. The sensitivity of SAA was better than both CRP and IL-6 from the initial presentation of infection until 24 hours after onset (95% versus 32% and 78% at 0 hours; 100% versus 53% and 47% at 8 hours; and 97% versus 84% and 19% at 24 hours, respectively). The specificity of SAA (93%) appeared to be comparable with CRP (97%) at 0 hours. The results indicate that while CRP and IL-6 are more specific than SAA, their sensitivities are much worse at different phases of the infection (i.e., CRP has unacceptably poor sensitivity during the early phase of infection, while IL-6 has unacceptably low sensitivity by 8 hours after onset). Thus, SAA is a more reliable marker throughout the first 24 hours after the onset of infection. The same team of investigators also studied the prognostic value of SAA in infected preterm infants.47 There were two study groups to which subjects were allocated: the non-fulminant sepsis group (infants who survived more than 72 hours) and the fulminant sepsis group (infants who survived less than 72 hours). Circulating concentrations of SAA, CRP and white blood cells were significantly lower in the fulminant than the non-fulminant group at 8, 24 and 48 hours after the onset of infection. Mortality was found to be significantly and inversely associated with SAA concentrations at 8 and 24 hours.47 This suggests that SAA concentrations can provide important prognostic information in infected infants as early as 8 hours after clinical presentation. Lipopolysaccharide-binding protein Lipopolysaccharidebinding protein (LBP) is an acute phase reactant similar to CRP and SAA, and is produced mainly by the liver.48 It binds to the lipopolysaccharide of bacteria to form a

BIOCHEMICAL MARKERS OF NEONATAL SEPSIS

complex which interacts with the macrophage receptor and initiates the pro-inflammatory host response.49 In view of the timing of this mechanism of action, investigators argue that LBP may be able to fill a ‘diagnostic gap’ between the early phase of infection, when IL-6 is useful, and the late phase, when CRP is up-regulated.48 LBP levels can peak earlier after onset of infection than CRP, 6 to 12 hours compared with 12 to 20 hours.42,48 In a study mixing subjects suffering from either early- or late-onset infection, LBP at the onset of infection showed superior sensitivity and negative predictive value (97 and 92%) compared with PCT (55 and 91%), IL-6 (55 and 91%), and CRP (70 and 94%, respectively).49 As the population comprised predominantly neonates less than 48 hours of age, the diagnostic utilities of PCT would be affected by its physiological fluctuation,39 while IL-6 and CRP could be affected by maternal and obstetrical factors.28 The investigators suggest that LBP probably has superior diagnostic utilities in early-onset neonatal infection than IL-6 and PCT. Another study focusing on early-onset infection in preterm and term infants also found that LBP response was significantly different between infected and non-infected infants by 12 hours after suspected onset of infection and was independent of gestational age.48 Thus far, the studies investigating the use of LBP in neonatal bacterial infection have been small with only rudimentary stratification into early or late-onset infection, preterm or term infants, and early or late phase of infection. Separate studies into each of these areas are required before the clinical application of this marker can be properly assessed. Components of the complement cascade The complement cascade can be activated via various inflammatory pathways and involves many different components. The classical pathway can be triggered by CRP and immune complexes, and the alternative pathway can be activated by lipopolysaccharide.50 These and other complement pathways may result in many products, of which C5a has been extensively studied recently.50,51 This complement-activation product is an anaphylatoxin that can induce proinflammatory changes, acting as a leukocyte chemoattractant and activating C5a receptor-positive cells, including neutrophils.50 Studies have indicated that increased levels of circulating C5a is associated with poor prognosis, multiorgan failure and death.50,52 Recently, a study investigating a second C5a receptor, neutrophil C5L2, has shown that decreased expression of this receptor on neutrophils is associated with worse prognosis.51 As C5L2 is not coupled with signalling G proteins and, therefore, does not induce pro-inflammatory responses when bound, it has been suggested to be a ‘scavenging’ receptor that may prevent excessive C5a activity. As most studies were performed in the adult population, more neonatal trials are needed to elucidate the prognostic value of this marker. Inter-a-inhibitor proteins There are high levels of inter-ainhibitor proteins (IaIp) in human plasma, a group of polypeptides that inhibit serine proteases and exhibit antiinflammatory properties.14 These proteins are thought to be involved in diverse pathophysiological processes, including the regulation of the host response to inflammation, tumour invasion and metastasis, wound healing and ovulation.53,54 An in vitro study shows no inhibitory effects

145

of even high concentrations of IaIp on bacterial growth, yet injecting mice with IaIp could increase the LD50 (dose at which there is 50% mortality) of a highly virulent strain of Escherichia coli by 100-fold.54 In the same report, it was shown that healthy adults (mean 872 + 234 mg/L) had higher circulating IaIp levels measured within 24 hours of onset of infection than septic subjects who survived more than 28 days (688 + 295 mg/L). Further, septic subjects who died before 28 days had significantly lower IaIp levels (486 + 193 mg/L) than either survivors or controls. Further analysis of the data demonstrated an inverse correlation between circulating IaIp concentrations and 28 day mortality in severely septic subjects. Baek et al. studied IaIp as a diagnostic marker in a group of preterm and term infants.14 Comparable with adults, high circulating concentrations of IaIp were also found in the newborn infants and the levels were not influenced by gestational age. Similarly, healthy infants (613 + 286 mg/L) were found to have significantly higher concentrations of IaIp than infected infants (169 + 126 mg/L). The investigators also found that maternal and newborn IaIp levels were not correlated, suggesting that the production mechanism for mother and baby is likely to be independent. Also, longitudinal measurements of infected infants showed normalisation of IaIp levels within 4–12 days of antimicrobial therapy, indicating a relatively wide window of opportunity for specimen sampling. Although the investigators suggest that IaIp may be helpful as a prognostic as well as a diagnostic marker in neonatal infection, the sample size was too small to establish a robust diagnostic model that could allow the clinician to confidently withhold antibiotic therapy in neonates suspected of bacterial infection.

OTHER MODALITIES OF INVESTIGATION IN NEONATAL INFECTION Apart from biochemical markers, recent research has also focused on other modalities of investigation of neonatal infection, including leukocyte surface antigens,55,56 and molecular techniques.57,58 Judicious selection of these investigative tools to be utilised in conjunction with current biochemical markers can increase diagnostic accuracy and aid the neonatal clinician in making decisions not only on when to commence or discontinue antibiotic therapy but also, in the case of molecular techniques, how to rationalise treatment at an early stage of disease. Leukocyte surface antigens Many membrane antigens are expressed on leukocyte cell surfaces. With the advent of flow cytometry, the densities of these surface antigens can be measured accurately and quickly using minute amounts of blood (50 mL).56 The pattern of leukocyte surface antigen expression can change in response to infection within minutes of exposure.13,56,59 The specific pattern of antigen expression on leukocytes has been thought to more accurately reflect the host immunological response to infection than the circulating concentrations of cytokines and acute phase reactants.9 CD64 is a high affinity receptor that binds the Fc portion of IgG antibodies found on monocytes and macrophages.59 It is usually expressed at very low densities on neutrophil

146

LAM AND NG

cell surfaces, but can be markedly up-regulated at the onset of infection.56,59 Therefore, neutrophil CD64 surface density can be used as an indicator for differentiating infected from non-infected infants at a very early stage. In a study of four leukocyte markers (neutrophil CD11b and CD64, and lymphocyte CD25 and CD45RO), neutrophil CD64 was found to be a sensitive and specific infection marker at the onset and up to 1 day after the initial presentation in both late-onset bacterial sepsis and necrotising enterocolitis (NEC) in VLBW infants.56 The sensitivity and specificity of neutrophil CD64 at 0, 24 and 48 hours were 95 and 88%, 97 and 90%, and 86 and 86%, respectively. The sensitivity of neutrophil CD64 compared favourably against the corresponding parameters of CD11b (70, 25 and 24%), IL-6 (78, 44 and 46%) and CRP (65, 72 and 65%). A large study using neutrophil CD64 as a marker of early-onset neonatal infection also revealed promising diagnostic utilities at the onset of infection (sensitivity 79%, specificity 89%) and 24 hours afterwards (sensitivity 96%, specificity 81%).59 In the same study, combining CRP with CD64 did not lead to substantially improved diagnostic utilities, but in a separate study on late-onset neonatal infection, combinations of CD64 with CRP or IL-6 could increase the sensitivity from 95 to 100%, and maintain a specificity approaching 90%.56 Despite a rapid increase in neutrophil CD64 expression at the onset of infection, unlike many other early phase markers including IL-6 and IL-8, the window of opportunity for sampling is much wider (424 hours), thus facilitating clinical use. CD11b is a b2-integrin adhesion molecule which is present on leukocyte cell surfaces at low levels when not in an activated state.13 Pro-inflammatory cytokines and chemokines, such as IL-8 can rapidly increase CD11b expression within minutes.9,55 With CD18, CD11b constitutes CR3 (CD11b/CD18), which bind molecules such as lipopolysaccharide and a component of the complement system, iC3b, to promote phagocytosis and lipopolysaccharide clearance.60 In view of the rapid response to infection, it has been thought to be useful in the very early phases of infection, where the diagnostic utilities of other markers such as PCT and CRP may not be sufficient. A team of investigators utilised this property of CD11b as an infection surveillance tool.13 Neutrophil CD11b cell surface densities were measured daily in a cohort of preterm infants regardless of symptoms. They found that laboratory evidence of infection based on neutrophil and monocyte CD11b could precede clinical suspicion of infection by up to 3 days. However, the disadvantages of CD11b include inter-centre variability,13,55,56,61 elevated densities secondary to non-infective causes such as respiratory distress syndrome,61 and unsatisfactory diagnostic utilities in lateonset neonatal infection and necrotising enterocolitis.56 Molecular techniques A major disadvantage of conventional microbiological culture of pathogens is the prolonged period required for incubation and identification of isolates, which may take up to 48–72 hours for clinically useful results. Pre-treatment with antibiotics and technical difficulties in obtaining adequate specimens for culture in neonates exacerbate this problem. An exciting new area of research in the field of

Pathology (2008), 40(2), February

neonatal infection is the application of molecular techniques, such as fluorescence in situ hybridisation (FISH),58 broad range polymerase chain reaction (PCR)62 and mRNA detection,63 to augment or even substitute such conventional investigations. Recently, by using FISH, investigators have been able to substantially reduce the time required to identify organisms isolated in blood cultures.58 It has been demonstrated that a reduction of more than 18 hours can be possible for bacterial isolates, and 42 hours for yeasts. A microarraybased technique to detect bacterial antibiotic resistance genes64,65 is another recent development that promises to significantly shorten the time required to generate clinically useful antibiotic sensitivity profiles. Slow growing and fastidious pathogens also pose a challenge to blood culturebased investigations. Direct detection of genetic material of pathogens by techniques such as broad range PCR62 have been employed to identify organisms in tissue fluids such as the pleural fluid.66 These techniques will need to be further refined, developed and studied in the neonatal population before widespread clinical use can be warranted. As the circulating concentrations of cytokines and chemokines may not fully reflect the host response to the infection,9 investigators have focused on peripheral blood cell biosynthesis of chemokine mRNA and its relation with perinatal infections and birth asphyxia.63 Whole blood IL-8 mRNA concentrations were found to increase in cases with either perinatal infections or perinatal asphyxia, whilst MCP-1 mRNA concentrations were only increased in cases with perinatal asphyxia. We predict that further studies of inflammatory mediator-associated mRNA production during different phases of early- and late-onset neonatal infection will give further insight into the complexities of the infection process and improve the diagnostic accuracy of biochemical markers of infection. Advances in proteomic technology67 have provided investigators with a new dimension in searching for diagnostic markers of neonatal infection. Thus far, investigators have used the ‘candidate’ approach by testing known proteins, mainly from the inflammatory cascade, for their diagnostic utilities. The proteomic techniques provide clinicians the alternative of the ‘hypothesis-free’ approach to search for novel protein markers. A few years ago, our research team embarked on the proteomic project to search for new infection markers in neonatal infection and necrotising enterocolitis. We have identified at least three proteins with differing molecular masses as potential candidates. Pending protein identification, we may hopefully utilise this new technique to discover new markers with novel properties for diagnosis and prognostication of infection or other conditions in newborn infants.

COST VERSUS BENEFIT OF INFECTION MARKERS TO MICROBIOLOGICAL DIAGNOSTIC TECHNIQUES Laboratory determination of new biomarkers of infection, such as leukocyte surface antigens, novel cytokines and chemokines, has often been costly and labour intensive before the advent of automation of sample processing and marker determination. However, conventional microbiological culture has major drawbacks, including poor sensitivity (e.g., inadequate specimen, patient pre-treated with

BIOCHEMICAL MARKERS OF NEONATAL SEPSIS

antimicrobial agents or fastidious and/or slow-growing organisms), lengthy turnaround time and difficulty in distinguishing contaminant organisms from genuine infecting pathogens (e.g., coagulase-negative Staphylococci or Bacillus species).9,10 This often leads to prolonged courses of antibiotics for non-infected patients and repeated courses of treatment for patients with conditions which may mimic infection, such as exacerbations of bronchopulmonary dysplasia, apnoea of prematurity or unstable temperature due to environmental factors. Not only does prolonged stay in the neonatal intensive care unit lead to substantial increase in expenditure (the cost of neonatal intensive care per patient per day in Hong Kong is approximately US$2000), the psychosocial disadvantages of keeping the parents and infant separated must also be considered. More seriously, cases presumed to be suffering from necrotising enterocolitis may result in prolonged total parenteral nutrition, and consequently repeated catheter related infection, cholestatic jaundice and trace element or nutritional deficiencies. Widespread use of broad spectrum antibiotics further increases the risk of drug-related complications, intravenous access associated morbidity, and emergence of multidrug-resistant organisms.

SUMMARY Compared with conventional haematological and microbiological techniques, newer infection markers have been shown to improve management of infants suspected to suffer from infection. Although the sensitivity of acute phase reactants such as CRP during the early phase of infection is relatively low (40–60%), its performance improves during the late phases (80–90%),19,68 and thus allows the neonatal clinician to stop antibiotics early in infants with no other evidence of infection. However, the characteristics of acute phase proteins in the early phase of infection imply that neonatologists are not able to rely on these markers to withhold treatment in infants with suspected clinical sepsis. Utilising early phase markers with good diagnostic utilities such as IL-6, IL-8, IP-10 and neutrophil CD64 can increase the confidence of neonatologists in reaching a definitive decision regarding antibiotic usage, as demonstrated in the multicentre trial using IL-8 and CRP.11 The application of flow cytometric technology in the field of neonatal infection has been a major breakthrough. By measuring cytokines, chemokines and cell surface antigens by cytometric bead array, multiple markers can be determined rapidly with tiny amounts of blood on an ad hoc basis. Prognostic markers such as IaIp, C5a, RANTES and proinflammatory:anti-inflammatory cytokine ratios (e.g., IL6:IL-10 ratio) can also provide neonatologists with crucial information about the severity of the clinical condition and the likely progress of the disease. Being able to predict the development of severe and potentially fatal infection from the outset can assist neonatal clinicians target a specific group of critically ill, infected infants for early and effective treatment with the possibility of improving morbidity and mortality. To date, the most practical and effective strategy is to incorporate early phase (e.g., CD64, IL-6, IL-8 and IP-10), late phase (e.g., CRP or SAA) and prognostic markers (e.g., IL-10 or RANTES) into a panel, and to measure this

147

combination of markers serially in order to monitor the immunological response to infection and treatment. Address for correspondence: Professor P. C. Ng, Department of Paediatrics, 6/F, Clinical Sciences Building, Prince of Wales Hospital, Sha Tin, New Territories, Hong Kong. E-mail: [email protected]

References 1. Stoll BJ, Hansen N, Fanaroff AA, et al. Late-onset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research Network. Pediatrics 2002; 110: 285–91. 2. Stoll BJ, Hansen N, Fanaroff AA, et al. Changes in pathogens causing early-onset sepsis in very-low-birth-weight infants. N Engl J Med 2002; 347: 240–7. 3. Philip AG. The evolution of neonatology. Pediatr Res 2005; 58: 799–815. 4. 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: 209–13. 5. van Dissel JT, van Langevelde P, Westendorp RG, Kwappenberg K, Frolich M. Anti-inflammatory cytokine profile and mortality in febrile patients. Lancet 1998; 351: 950–3. 6. Schultz C, Temming P, Bucsky P, Gopel W, Strunk T, Hartel C. Immature anti-inflammatory response in neonates. Clin Exp Immunol 2004; 135: 130–6. 7. Stoll BJ, Gordon T, Korones SB, et al. Early-onset sepsis in very low birth weight neonates: a report from the National Institute of Child Health and Human Development Neonatal Research Network. J Pediatr 1996; 129: 72–80. 8. Stoll BJ, Gordon T, Korones SB, et al. Late-onset sepsis in very low birth weight neonates: a report from the National Institute of Child Health and Human Development Neonatal Research Network. J Pediatr 1996; 129: 63–71. 9. Ng PC. Diagnostic markers of infection in neonates. Arch Dis Child Fetal Neonatal Ed 2004; 89: 229–35. 10. Polin RA. The ‘ins and outs’ of neonatal sepsis. J Pediatr 2003; 143: 3–4. 11. Franz AR, Bauer K, Schalk A, et al. Measurement of interleukin 8 in combination with C-reactive protein reduced unnecessary antibiotic therapy in newborn infants: a multicenter, randomized, controlled trial. Pediatrics 2004; 114: 1–8. 12. Horisberger T, Harbarth S, Nadal D, Baenziger O, Fischer JE. G-CSF and IL-8 for early diagnosis of sepsis in neonates and critically ill children - safety and cost effectiveness of a new laboratory prediction model: study protocol of a randomized controlled trial. Crit Care 2004; 8: 443–50. 13. Turunen R, Andersson S, Nupponen I, Kautiainen H, Siitonen S, Repo H. Increased CD11b-density on circulating phagocytes as an early sign of late-onset sepsis in extremely low-birth-weight infants. Pediatr Res 2005; 57: 270–5. 14. Baek YW, Brokat S, Padbury JF, Pinar H, Hixson DC, Lim YP. Interalpha inhibitor proteins in infants and decreased levels in neonatal sepsis. J Pediatr 2003; 143: 11–5. 15. 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: 462–8. 16. Verboon-Maciolek MA, Thijsen SF, Hemels MA, et al. Inflammatory mediators for the diagnosis and treatment of sepsis in early infancy. Pediatr Res 2006; 59: 457–61. 17. Lehrnbecher T, Schrod L, Kraus D, Roos T, Martius J, von Stockhausen HB. Interleukin-6 and soluble interleukin-6 receptor in cord blood in the diagnosis of early onset sepsis in neonates. Acta Paediatr 1995; 84: 806–8. 18. Ng PC, Li K, Leung TF, et al. Early prediction of sepsis-induced disseminated intravascular coagulation with interleukin-10, interleukin6, and RANTES in preterm infants. Clin Chem 2006; 52: 1181–9. 19. Ng PC, Cheng SH, Chui KM, et al. Diagnosis of late onset neonatal sepsis with cytokines, adhesion molecule, and C-reactive protein in preterm very low birthweight infants. Arch Dis Child Fetal Neonatal Ed 1997; 77: 221–7. 20. Procianoy RS, Silveira RC. The role of sample collection timing on interleukin-6 levels in early-onset neonatal sepsis. J Pediatr (Rio J) 2004; 80: 407–10. 21. Resch B, Gusenleitner W, Muller W. Procalcitonin, interleukin-6, C-reactive protein and leukocyte counts in infants with bronchiolitis. Pediatr Infect Dis J 2003; 22: 475–6.

148

LAM AND NG

22. Resch B, Gusenleitner W, Muller WD. Procalcitonin and interleukin-6 in the diagnosis of early-onset sepsis of the neonate. Acta Paediatr 2003; 92: 243–5. 23. Chiesa C, Pellegrini G, Panero A, et al. C-reactive protein, interleukin6, and procalcitonin in the immediate postnatal period: influence of illness severity, risk status, antenatal and perinatal complications, and infection. Clin Chem 2003; 49: 60–8. 24. Silveira RC, Procianoy RS. Evaluation of interleukin-6, tumour necrosis factor-alpha and interleukin-1beta for early diagnosis of neonatal sepsis. Acta Paediatr 1999; 88: 647–50. 25. Kishimoto T. The biology of interleukin-6. Blood 1989; 74: 1–10. 26. Nishimoto N, Kishimoto T. Interleukin 6: from bench to bedside. Nat Clin Pract Rheumatol 2006; 2: 619–26. 27. Hodge G, Hodge S, Haslam R, et al. Rapid simultaneous measurement of multiple cytokines using 100 microl sample volumes – association with neonatal sepsis. Clin Exp Immunol 2004; 137: 402–7. 28. D’Alquen D, Kramer BW, Seidenspinner S, et al. Activation of umbilical cord endothelial cells and fetal inflammatory response in preterm infants with chorioamnionitis and funisitis. Pediatr Res 2005; 57: 263–9. 29. Ng PC, Lam HS. Diagnostic markers for neonatal sepsis. Curr Opin Pediatr 2006; 18: 125–31. 30. Schultz C, Strunk T, Temming P, Matzke N, Hartel C. Reduced IL-10 production and -receptor expression in neonatal T lymphocytes. Acta Paediatr 2007; 96: 1122–5. 31. Osuchowski MF, Welch K, Siddiqui J, Remick DG. Circulating cytokine/inhibitor profiles reshape the understanding of the SIRS/ CARS continuum in sepsis and predict mortality. J Immunol 2006; 177: 1967–74. 32. Dembinski J, Behrendt D, Heep A, Dorn C, Reinsberg J, Bartmann P. Cell-associated interleukin-8 in cord blood of term and preterm infants. Clin Diagn Lab Immunol 2002; 9: 320–3. 33. Franz AR, Sieber S, Pohlandt F, Kron M, Steinbach G. Whole blood interleukin 8 and plasma interleukin 8 levels in newborn infants with suspected bacterial infection. Acta Paediatr 2004; 93: 648–53. 34. Reinsberg J, Dembinski J, Dorn C, Behrendt D, Bartmann P, van Der Ven H. Determination of total interleukin-8 in whole blood after cell lysis. Clin Chem 2000; 46: 1387–94. 35. Orlikowsky TW, Neunhoeffer F, Goelz R, et al. Evaluation of IL-8concentrations in plasma and lysed EDTA-blood in healthy neonates and those with suspected early onset bacterial infection. Pediatr Res 2004; 56: 804–9. 36. Fotopoulos S, Mouchtouri A, Xanthou G, Lipsou N, Petrakou E, Xanthou M. Inflammatory chemokine expression in the peripheral blood of neonates with perinatal asphyxia and perinatal or nosocomial infections. Acta Paediatr 2005; 94: 800–6. 37. Mackay CR. Chemokines: immunology’s high impact factors. Nat Immunol 2001; 2: 95–101. 38. Ng PC, Li K, Chui KM, et al. IP-10 is an early diagnostic marker for identification of late-onset bacterial infection in preterm infants. Pediatr Res 2007; 61: 93–8. 39. Turner D, Hammerman C, Rudensky B, Schlesinger Y, Goia C, Schimmel MS. Procalcitonin in preterm infants during the first few days of life: introducing an age related nomogram. Arch Dis Child Fetal Neonatal Ed 2006; 91: 283–6. 40. van Rossum AM, Wulkan RW, Oudesluys-Murphy AM. Procalcitonin as an early marker of infection in neonates and children. Lancet Infect Dis 2004; 4: 620–30. 41. Arnon S, Litmanovitz I, Regev R, et al. Serum amyloid A protein is a useful inflammatory marker during late-onset sepsis in preterm infants. Biol Neonate 2005; 87: 105–10. 42. Ehl S, Gehring B, Pohlandt F. A detailed analysis of changes in serum C-reactive protein levels in neonates treated for bacterial infection. Eur J Pediatr 1999; 158: 238–42. 43. Chiesa C, Panero A, Rossi N, et al. Reliability of procalcitonin concentrations for the diagnosis of sepsis in critically ill neonates. Clin Infect Dis 1998; 26: 664–72. 44. Joram N, Boscher C, Denizot S, et al. Umbilical cord blood procalcitonin and C reactive protein concentrations as markers for early diagnosis of very early onset neonatal infection. Arch Dis Child Fetal Neonatal Ed 2006; 91: 65–6. 45. He R, Sang H, Ye RD. Serum amyloid A induces IL-8 secretion through a G protein-coupled receptor, FPRL1/LXA4R. Blood 2003; 101: 1572–81.

Pathology (2008), 40(2), February

46. Lannergard A, Friman G, Ewald U, Lind L, Larsson A. Serum amyloid A (SAA) protein and high-sensitivity C-reactive protein (hsCRP) in healthy newborn infants and healthy young through elderly adults. Acta Paediatr 2005; 94: 1198–1202. 47. Arnon S, Litmanovitz I, Regev R, Lis M, Shainkin-Kestenbaum R, Dolfin T. The prognostic virtue of inflammatory markers during lateonset sepsis in preterm infants. J Perinat Med 2004; 32: 176–80. 48. Behrendt D, Dembinski J, Heep A, Bartmann P. Lipopolysaccharide binding protein in preterm infants. Arch Dis Child Fetal Neonatal Ed 2004; 89: 551–4. 49. Pavcnik-Arnol M, Hojker S, Derganc M. Lipopolysaccharide-binding protein in critically ill neonates and children with suspected infection: comparison with procalcitonin, interleukin-6, and C-reactive protein. Intensive Care Med 2004; 30: 1454–60. 50. Ward PA. The dark side of C5a in sepsis. Nat Rev Immunol 2004; 4: 133–42. 51. Huber-Lang M, Sarma JV, Rittirsch D, et al. Changes in the novel orphan, C5a receptor (C5L2), during experimental sepsis and sepsis in humans. J Immunol 2005; 174: 1104–10. 52. Riedemann NC, Guo RF, Ward PA. A key role of C5a/C5aR activation for the development of sepsis. J Leukoc Biol 2003; 74: 966–70. 53. Fries E, Kaczmarczyk A. Inter-alpha-inhibitor, hyaluronan and inflammation. Acta Biochim Pol 2003; 50: 735–42. 54. Lim YP, Bendelja K, Opal SM, Siryaporn E, Hixson DC, Palardy JE. Correlation between mortality and the levels of inter-alpha inhibitors in the plasma of patients with severe sepsis. J Infect Dis 2003; 188: 919–26. 55. Nupponen I, Andersson S, Jarvenpaa AL, Kautiainen H, Repo H. Neutrophil CD11b expression and circulating interleukin-8 as diagnostic markers for early-onset neonatal sepsis. Pediatrics 2001; 108: E12. 56. Ng PC, Li K, Wong RP, Chui KM, Wong E, Fok TF. Neutrophil CD64 expression: a sensitive diagnostic marker for late-onset nosocomial infection in very low birthweight infants. Pediatr Res 2002; 51: 296–303. 57. Peters RP, van Agtmael MA, Danner SA, Savelkoul PH, Vandenbroucke-Grauls CM. New developments in the diagnosis of bloodstream infections. Lancet Infect Dis 2004; 4: 751–60. 58. Peters RP, Savelkoul PH, Simoons-Smit AM, Danner SA, Vandenbroucke-Grauls CM, van Agtmael MA. Faster identification of pathogens in positive blood cultures by fluorescence in situ hybridization in routine practice. J Clin Microbiol 2006; 44: 119–23. 59. Ng PC, Li G, Chui KM, et al. Neutrophil CD64 is a sensitive diagnostic marker for early-onset neonatal infection. Pediatr Res 2004; 56: 796–803. 60. Gasque P. Complement: a unique innate immune sensor for danger signals. Mol Immunol 2004; 41: 1089–98. 61. Nupponen I, Pesonen E, Andersson S, et al. Neutrophil activation in preterm infants who have respiratory distress syndrome. Pediatrics 2002; 110: 36–41. 62. Rantakokko-Jalava K, Jalava J. Optimal DNA isolation method for detection of bacteria in clinical specimens by broad-range PCR. J Clin Microbiol 2002; 40: 4211–7. 63. Petrakou E, Mouchtouri A, Levi E, Lipsou N, Xanthou M, Fotopoulos S. Interleukin-8 and monocyte chemotactic protein-1 mRNA expression in perinatally infected and asphyxiated preterm neonates. Neonatology 2007; 91: 107–13. 64. Perreten V, Vorlet-Fawer L, Slickers P, Ehricht R, Kuhnert P, Frey J. Microarray-based detection of 90 antibiotic resistance genes of grampositive bacteria. J Clin Microbiol 2005; 43: 2291–2302. 65. Call DR, Bakko MK, Krug MJ, Roberts MC. Identifying antimicrobial resistance genes with DNA microarrays. Antimicrob Agents Chemother 2003; 47: 3290–5. 66. Saglani S, Harris KA, Wallis C, Hartley JC. Empyema: the use of broad range 16S rDNA PCR for pathogen detection. Arch Dis Child 2005; 90: 70–3. 67. Hortin GL, Jortani SA, Ritchie JC Jr, Valdes R Jr, Chan DW. Proteomics: a new diagnostic frontier. Clin Chem 2006; 52: 1218–22. 68. Benitz WE, Han MY, Madan A, Ramachandra P. Serial serum C-reactive protein levels in the diagnosis of neonatal infection. Pediatrics 1998; 102: E41.