International Journal of Infectious Diseases 30 (2015) 27–32

Contents lists available at ScienceDirect

International Journal of Infectious Diseases journal homepage: www.elsevier.com/locate/ijid

Can sTREM-1 predict septic shock & death in late-onset neonatal sepsis? A pilot study§ Vı´ctor Arı´zaga-Ballesteros a, Mario Rene´ Alcorta-Garcı´a a, Lizzeth Carolina La´zaro-Martı´nez a, Jesu´s Manuel Ame´zquita-Go´mez a, Jose´ Manuel Alanı´s-Cajero a, Luis Villela b, Fabiola Castorena-Torres b, Vı´ctor Javier Lara-Dı´az a,b,* a b

Tecnolo´gico de Monterrey, Escuela de Medicina. Programa Multice´ntrico de Residencias Me´dicas, Neonatologı´a Tecnolo´gico de Monterrey, Escuela de Medicina, Direccio´n de Investigacio´n e Innovacio´n

A R T I C L E I N F O

Article history: Received 16 July 2014 Received in revised form 28 October 2014 Accepted 30 October 2014 Keywords: Premature Diseases Intensive care Neonatal Sepsis Diagnosis ROC curve Sensitivity and Specificity.

S U M M A R Y

Background and Objectives: The transmembrane glycoprotein TREM-1 triggers an inflammatory response. Its soluble fraction (sTREM-1) has been shown to have diagnostic accuracy for late-onset neonatal sepsis (LONS). Until now, the potential of sTREM-1 to predict septic shock and/or death in septic neonates has not been explored. This study obtained estimates of the incidence and prevalence of septic shock and/or death in septic neonates for future sample size calculations for confirmatory studies and evaluated the feasibility of using sTREM-1 as a predictor of septic shock and/or death in neonates with LONS criteria. Study design: A pilot study with a cross-sectional design was performed from May 1st to October 31st, 2012. The participants were hospitalized neonates who, after three days of life, were diagnosed as having LONS. Plasma sTREM-1 was quantified by ELISA. The main outcome measurement was the development of septic shock and/or death. Results: Of 71 eligible subjects, nine (12.7%) progressed to septic shock and/or death. In the LONS-NonShock group, the sTREM-1 median and interquartile range (IQR) plasma value were 10 (10 to 70) pg/mL. In the LONS & Shock/Death group, the values were 567 (260 to 649) pg/mL. These values were significantly different (Mann-Whitney’s U test, p = 0.001). A ROC curve for a proposed sTREM-1 cut-off value of 300 pg/mL exhibited an area under the curve of 0.884 (95% CI = 0.73 to 1.0; p < 0.0001), with a sensitivity of 0.78 (95% CI = 0.46 to 0.94) and specificity of 0.97 (95% CI = 0.92 to 0.99); PPV would be 0.78 (95% CI = 0.46 to 0.94) and NPV 0.97 (95% CI = 0.92 to 0.99). Conclusions: In neonates with LONS, sTREM-1 has the potential to provide an excellent predictive value for septic shock/death. Larger sample sizes are needed to identify the optimal cut-off value of plasma sTREM-1 for this diagnosis and to provide diagnostic accuracy measures. ß 2014 The Authors. Published by Elsevier Ltd on behalf of International Society for Infectious Diseases. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/bync-sa/3.0/).

1. Introduction The transmembrane glycoprotein TREM-1, a 30 kDa member of the immunoglobulin superfamily, triggers an inflammatory

§ The first draft and subsequent revisions were written by Dr. Vı´ctor Javier LaraDı´az. No honorarium, grant or other form of payment was provided for the production of the manuscript. * Corresponding author. CITES, Centro de Innovacio´n y Transferencia en Salud, Ave. Ignacio Morones Prieto 3000 poniente, Tercer piso, Oficina 313. Col. Doctores, CP 64710, Phone +52 81 8888-2147, Fax: +52 81 8888-2052; Monterrey, Nuevo Leo´n, Me´xico. E-mail addresses: [email protected], [email protected] (V.J. Lara-Dı´az).

response by releasing cytokines, increasing cell surface receptors, and activating neutrophil degranulation and oxidative burst.1,2 The soluble fraction of sTREM-1 has been studied as a predictive tool for sepsis, septic shock, and death in adults.3–7 The morbidity and mortality rates of nosocomial or late-onset neonatal sepsis (LONS) cases remain high, especially in preterm neonates.8–10 Until now, the potential of sTREM-1 to specifically predict septic shock and death in septic neonates has not been evaluated. Therefore, we obtained estimates of the incidence of LONS and the prevalence of septic shock for future sample size calculations and assessed the feasibility of sTREM-1 as a predictor of these specific outcomes in this population.

http://dx.doi.org/10.1016/j.ijid.2014.10.013 1201-9712/ß 2014 The Authors. Published by Elsevier Ltd on behalf of International Society for Infectious Diseases. This is an open access article under the CC BY-NC-SA license (http://creativecommons.org/licenses/by-nc-sa/3.0/).

28

V. Arı´zaga-Ballesteros et al. / International Journal of Infectious Diseases 30 (2015) 27–32

2. Patients and Methods This study was conducted from May 1st to October 31st, 2012, in two medical centers located in Northeastern Mexico: The Hospital Regional Materno-Infantil (Center 1) and The Hospital Metropolitano Dr. Bernardo Sepu´lveda (Center 2). These two public hospitals have level II-III neonatal intensive care units and belong to the Servicios de Salud Network in Nuevo Leo´n, Mexico. The inclusion criteria for this study population were as follows: inborn neonates; suspicion of LONS, defined as an infection occurring after 72 hours of life; a systemic inflammatory response syndrome based on the criteria of the International Pediatric Sepsis Consensus Conference and Haque10,11; and a nosocomial sepsis predictive score (NOSEP-1) above 8 points. The NOSEP-1 is determined based on C-reactive protein level, neutrophil fraction, thrombocytopenia, fever, and prolonged parenteral nutrition. The NOSEP-1 score provides a valuable tool for the early identification of nosocomial sepsis12,13 and has been validated in our population.14 Because this was an exploratory study, a sequential, nonprobabilistic sampling method was employed, and no prior sample size estimation was performed. Neonates born from mothers with clinical chorioamnionitis, neonates born from HIV positive mothers and neonates who had early-onset sepsis, congenital infections, congenital anomalies, endocrine disorders, dehydration, asphyxia or birth trauma were excluded. The study protocol was approved by the institutional human research review boards at Escuela de Medicina, Tecnolo´gico de Monterrey, and by the corresponding review boards at both study hospitals, with the identifier VPsTREMP/12/480. Eligible neonates were included once clinical derangement attributed to LONS occurred, and a written informed consent was obtained from their parents or legal guardians. The data recorded included gestational age, birth weight, gender, method of delivery, age at diagnosis of LONS, age at development of main outcome, septic shock and survival. The included neonates were followed up for seven days after inclusion to observe the appearance of septic shock and/or death. Attributable mortality was defined as death occurring within seven days of inclusion in the study. The enrolled subjects were further analyzed according to their progression to study outcomes, shock and/or death (LONS & Shock/Death group), and those who did not develop shock and survived (LONS-Non Shock group). All participants, irrespective of the results of sTREM-1 determination, were included in the analysis. 2.1. Biochemical measurements The initial laboratory tests included a differential white blood cell count and platelet count, measurement of the C-reactive protein (CRP) level, blood culture and a sTREM-1 measurement in plasma obtained from a blood sample. Plasma was separated from blood and cryopreserved at -80 8C until analysis. Plasma sTREM-1 was quantified with the STREM Human ELISA kit (Cat. ab100659, Abcam, USA) by a certified laboratory professional who was blinded to the subject’s clinical status. The minimum limit of detection (LOD) of sTREM-1 with this method was 20 pg/mL. The results of the sTREM-1 measurement were not available to attending clinicians; all samples were analyzed after study recruitment was complete. Septic shock was considered to have occurred when neonates with LONS had hemodynamic deterioration defined as tachycardia (heart rate above 180 beats/min) and mean arterial blood pressure < 30 mmHg in two consecutive measurements two hours apart that required fluid and inotropic therapy. The diagnosis of septic shock also depended on two or more of the following criteria: capillary refill time > 4 seconds; oliguria, with

diuresis < 0.5 mL/h; and metabolic acidosis with a lactic acid level above 2 mmol/L.10,11 Laboratory evidence of infection was a CRP >14 mg/L, neutrophilia >50% and thrombocytopenia 38 8C and total parenteral nutrition for 14 days, as are defined in the NOSEP-1 score.12,13 All data were de-identified, and participants were referred to only by case numbers. 2.2. Statistical analysis Data were not normally distributed, therefore non-parametric statistical tools were used for the analysis. Values are expressed as medians and interquartile ranges (IQR); categorical data are expressed as frequencies and proportions. Mann-Whitney U, x2 or Fisher’s exact tests were used for comparisons as appropriate. Receiver operating characteristic curves (ROC) were constructed to compare neonates affected or non-affected by septic shock. Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) were calculated based on cutoff values derived from the ROC analysis. Values with a p < 0.05 were considered statistically significant. The statistical analysis was performed using IBM-SPSS Statistics 19 software (IBM Corporation, Armonk, NY, USA). 3. Results Over six months, a total of 11,629 live births occurred at the study hospitals. The patient flow and eligibility process are depicted in Figure 1. Overall, 850 (7.3%) live neonates were admitted to intensive care nurseries. Seventy-five of those admissions met the established criteria of LONS and thus were eligible for the study. From these, 71 were included. The demographic characteristics of the entire study population are shown in Table 1. The incidence of LONS during the study period was 88 per 1000 NICU admissions (95% CI = 70 to 109). There were slightly higher proportions of males and neonates born by cesarean section. A majority of subjects had a gestational age equal to or lower than 37 weeks and a birth weight below 2500 g. For the 71 subjects included in the study, the sTREM-1 values were below LOD in 43 subjects (60.6%) and above LOD in the remaining

8375 live births, Center 1

3254 live births, Center 2

850 admitted to NICU

75 eligible 0 declined to participate 4 unable to obtain samples 71 included (100%)

LONS & Shock/Death 9 (12.7%)

sTREM-1 LOD 8

LONS Non-Shock 62 (87.3%)

sTREM-1 LOD 20

Figure 1. Legend: Patient flow and eligibility chart. Center 1: Hospital Regional Materno-Infantil, SSNL; Center 2: Hospital Metropolitano Dr. Bernardo Sepu´lveda, SSNL. Patients recruited from May 1st to October 31, 2012. Patients with LONS criteria included all gestational age and birth weight categories. Eligible patients were those with LONS criteria and inclusion criteria. NICU = Newborn intensive care unit; sTREM-1 = soluble fraction of triggering receptor expressed on myeloid cells1; LOD = Limit of detection; LONS = Late-onset neonatal sepsis.

V. Arı´zaga-Ballesteros et al. / International Journal of Infectious Diseases 30 (2015) 27–32

29

Table 1 Demographic characteristics of the study group, n = 71 Value

Characteristic a

Hospital Center 1 Center 2 Gender a Male Female Method of delivery a Vaginal Abdominal Gestational age, weeks.b Birth weight, g.b Age at inclusion, days.b NOSEP-1 score [refs 13–15].b

39 (55) 32 (45) 37 (52) 34 (48) 32 39 33 1750 10 13

(45) (55) (30–37) (1076–2480) (5–25) (9–14)

Legend: Center 1: Hospital Materno-Infantil de Alta Especialidad, SSNL; Center 2: Hospital Metropolitano Dr. Bernardo Sepu´lveda, SSNL; a Frequency (%); b Median (interquartile range). References 13–15 abound on NOSEP-1 score.

Figure 2. Legend: Box and whisker plot of sTREM-1 concentrations (pg/mL) in the LONS-Non-Shock (empty box) and the LONS & Shock/Death (vertical bars box) groups.8 and * indicate atypical values. Mann-Whitney U test, p < 0.001.

28 subjects (39.4%); the sTREM-1 values were above LOD in 20 subjects in the LONS-Non Shock group and in 8 of 9 subjects in the LONS & Shock/Death group. Of the 71 enrolled neonates, nine (12.7%) progressed to septic shock and/or death. Thus, in our study population, the septic shock and/or death prevalence was 10.6 per 1000 NICU admissions (95% CI = 4.9 to 20). In those subjects in whom septic shock developed, it occurred within a median of 1.0 days (IQR = 1 to1.5) after study inclusion. Five deaths occurred (7%); all of them were within two days after study inclusion, and four of them were unequivocally preceded by septic shock; thus, the mortality attributable to septic shock was 4.7 per 1000 NICU admissions. Those who died did so at a median of 1.5 days (IQR = 1.13 to 2.0) after study inclusion. Considering all eligible subjects, death was strongly associated with shock (p = 0.006). Subjects who developed shock had an odds ratio for death of 23 (95% CI = 3 to 181). When considering only Table 2 Summary and comparison of data between groups.

Gender a Male Female Method of Delivery Vaginal Abdominal Hospital a Center 1 Center 2 Blood culture a Positive Negative Gestational age, Weeks b Birth weight, g b Age at inclusion, Days b NOSEP-1 score b sTREM-1, pg/mLb

LONS-Non Shock

LONS & Shock/Death

n = 62

n=9

p value

33 (53) 29 (47)

4 (44) 5 (56)

NS

c

28 (45) 34 (55)

4 (44) 5 (56)

NS

c

33 (53) 29 (47)

6 (67) 3 (33)

NS

c

3 (33) 6 (67) 30 (28–36)

NS NS

c

1260 (795–2680) 8 (5–9)

NS NS

d

those subjects who progressed towards septic shock and were positive for plasma sTREM-1, the association of shock with fatal outcome remained significant (p = 0.038), and the odds ratio for death was 15 (95% CI = 1.22 to 181). As stated in the methods section, all subjects, irrespective of the results of sTREM-1 determination below or above LOD, were included in the analysis and the estimation of diagnostic accuracy. Data for the comparison between groups are summarized and compared in Table 2. In our study population, sTREM-1 had a median of 10 pg/mL with an IQR = 10 to 95 pg/mL. A greatly significant difference for sTREM-1 values between the LONS-Non Shock and LONS & Shock/Death groups was found (p = 0.001) (Figure 2). The ROC curve of the measured values of sTREM-1 as a predictor of septic shock or death in this population yielded an area under the curve of 0.884 (95% CI = 0.735 to1.0; p < 0.0001) (Fig. 3). By examination of the coordinates of the ROC curve, we chose a potential sTREM-1 cut-off value of 300 pg/mL, which would result in the best relation of sensitivity to specificity. Since our goal was to make a prediction of the outcome on neonates affected with

a

23 (37) 39 (63) 34 (30–37) 1775 (1175–2483) 10.5 (5–28) 13 (9.8–14) 10 (10–70)

13 (8–13) 567 (260–649)

d

d

NS d 0.001

d

Legend: LONS = Late-onset neonatal sepsis; LOD = Limit of Detection; NA = Not applicable, since this category had only one subject; NS = Not significant; Center 1: Hospital Materno-Infantil de Alta Especialidad, SSNL; Center 2: Hospital Metropolitano Dr. Bernardo Sepu´lveda, SSNL. a Frequency (%); b Median (interquartile range); c Fisher’s exact test; d Mann-Whitney’s U test.

Figure 3. Legend: Receiver operating characteristic (ROC) curve for sTREM-1 values obtained from all participants, area under the ROC curve (AUC) 0.884 (95% CI = 0.735 to 1.0), p < 0.0001.

30

V. Arı´zaga-Ballesteros et al. / International Journal of Infectious Diseases 30 (2015) 27–32

Table 3 Frequency distribution of sTREM-1 results around the proposed cut-off value in each outcome group, n = 71. Outcome Group Value of sTREM-1, pg/mL

LONS-Non-Shocka

LONS &Shock/Deatha

Above 300 Below 300

2 (2.8) 60 (84.5)

7 (9.9) 2 (2.8)

Legend: LONS = Late onset neonatal sepsis; a Values are expressed as frequencies (percentage).

LONS with the results of a sTREM-1 determination, and in many instances the test yielded values below LOD, the real world calculation of diagnostic accuracy should include all participants, irrespective of the sTREM-1 result. Doing so, and grouping them above and below the proposed cut-off value, would yield the distribution shown in Table 3. Therefore, from data of the whole study population, a proposed sTREM-1 cut-off value of 300 pg/mL would exhibit a sensitivity of 0.78 (95% CI = 0.46 to 0.94) and specificity of 0.97 (95% CI = 0.92 to 0.99); PPV would be 0.78 (95% CI = 0.46 to 0.94) and NPV 0.97 (95% CI = 0.92 to 0.99). No correlation was observed for sTREM-1 and gestational age; birth weight; method of delivery; NOSEP-1 score; total leukocyte number; proportion of neutrophils, eosinophils, basophils or monocytes; platelet number or C-reactive protein level. 4. Discussion In this study period, our cumulative incidence of LONS was 88 per 1000 newborn intensive care unit (NICU) admissions, which was lower than the rate of 211 per 1000 VLBW neonate NICU admissions reported by Stoll BJ et al. in 2002 and substantially higher than the 6.3 episodes per 1000 admissions published by Cohen-Wolkowiez et al. in 2009. Furthermore, our findings are very similar to the 71/1000 NICU admissions reported by Vergnano in 2011 and the 74 per 1000 preterm neonate NICU admissions reported by Grisaru-Soen et al. from Israel in 2012.8,9,16–18 The variation in these reported rates may reflect differences in study methodologies, LONS definitions or even actual differences in the occurrence of these infections. Our incidence of septic shock, 10.6/1000 NICU admissions, was higher than the incidence of 2.2/1000 admissions reported by Kermorvant-Duchemin et al.9 Although no data are available on LONS and septic shock prevalence in underdeveloped countries, such as ours, it is reasonable that similar results may be obtained by future studies in underdeveloped countries. In 2010, Sarafidis et al. published the results of their prospective, observational study that aimed to compare the diagnostic accuracy for LONS of sTREM-1 values to those of interleukin-6. They selected 143 pg/mL of plasma sTREM-1 as the best value for detecting infected neonates; however, sTREM-1 was not found to be more accurate than IL-6 in differentiating infected from non-infected neonates.10 Our study of a population of LONS cases generated highly similar median and range values for plasma sTREM-1, 140 (70 to 518) pg/mL, to those reported by Sarafidis et al. A crucial difference between their work and ours is that we aimed to identify the subset of LONS neonates who were prone to an unfavorable outcome (shock or death). Mazzucchelli et al. recently published their findings for a cohort of 71 neonates, 16 of whom developed LONS and were compared to 16 non-infected neonates. They obtained a sTREM-1 median value of 49.5 pg/mL with an IQR of 13.5 to 176 pg/mL in infected neonates, which is lower than our values. Mazzucchelli et al. advocate the use of the CD64 index instead of sTREM-1, which in their series showed great heterogeneity; nevertheless, they recognize that their sample size

was small due to the nested case-control study design they employed.19 Adly A.A. et al. reported much higher sTREM-1 values than those of our study in a cohort of 112 infected neonates, but they included patients with both early-onset and late-onset sepsis. In their cohort, 75 participants can be identified as LONS, and in a subset of 39 of these with culture-proven infection, they report a mean  standard deviation of 1134.6  444 pg/mL. Overall, they proposed a cut-off level of 310 pg/mL for sTREM-1, as a sensitive and specific predictor of neonatal infection, which is in agreement with our findings in the sickest patients. They found much higher values of sTREM-1 in those individuals who died, 2053  889 pg/mL, than those who survived and affirmed that a sTREM-1 cutoff value of 1100 pg/mL is a good predictor of survival.20 The mean sTREM-1 values for the control subjects reported by their group were higher than those reported by Sarafidis and our team for infected neonates and were much higher than the results reported by Mazzucchelli et al.10,19 Interestingly, we had undetectable levels of sTREM-1 in roughly 61% of our subjects, even though all of them had strong clinical indicators of infection. Neither Sarafidis, Mazzucchelli or Adly reported how many subjects in their series had a sTREM-1 value below LOD.10,19,20 We found that the infants who progressed to septic shock had significantly higher sTREM-1 levels than those with sepsis who did not develop shock. Irrespective of different sTREM-1 values, which could be the result of different study populations and the ELISA kits used, previous works are in general agreement with our findings, in the sense that sTREM-1 levels were higher in the sickest neonates, and higher values are associated with poor prognosis.10,19,20 We speculate that the possibility of sTREM-1 differences in relation to genetic background should be explored in future trials. In our population of neonates with LONS, our data clearly suggest that a sTREM-1 value above 300 pg/mL accurately predicts a progression to septic shock/death. Garofoli et al., in a study of healthy neonates, established a significant negative correlation of sTREM-1 values and birth weight.21 Similar results were reported for the membrane-bound receptor by Matoba N et al. in neonates born preterm. Furthermore, they proposed a potential correlation of inflammation with preterm labor.22 Adly et al. also reported a significant positive sTREM-1 correlation with total leukocyte count, C-reactive protein and sepsis score.20 In our study, we could not confirm a significant positive or negative correlation of sTREM-1 with those variables, perhaps because we were focused on the evaluation of sTREM-1 in a group of neonates developing a complication of care (LONS) that does not occur at birth, but rather occurs at least three days after delivery. Instead, we observed that preterm neonates had a statistically significant lower median and interquartile range of sTREM-1 values than term neonates (data not shown). The attributable mortality in our study reached 7%, which is greater than that reported by Grisaru-Soen in 2012 in a similar population.18 The sickest newborns, those who had progressed to septic shock, accounted for four of the five observed deaths in our study. The expression of TREM-1 by effector cells has been shown to be dramatically increased in skin, biological fluids and tissues infected by Gram-positive and Gram-negative bacteria and fungi;2,5,7,23 however, in our study, 6 of 9 subjects with septic shock had negative blood cultures, even though they had the highest sTREM1 values of the study population and progressed to septic shock. Overall, 36.6% of the blood cultures in our series were positive. In previous works in the same hospitals, we reported a blood culture positivity rate in the vicinity of 42%.24 This lower rate of blood culture positivity was likely due to the technical difficulties in obtaining an adequate amount of blood for culture in smaller neonates. Schelonka et al. (1996) and Kellogg et al. (1997)

V. Arı´zaga-Ballesteros et al. / International Journal of Infectious Diseases 30 (2015) 27–32

published that a low level of bacteremia may occur in up to two thirds of the cases of neonatal sepsis, and with the typical amount of less than 4 to 10 bacterial CFU/mL in neonatal blood, volumes lower than 1-1.5 mL should be considered inadequate for a sensitive and timely detection of bacteremia.25,26 Because a tendency toward lower birth weight was observed among this group, although no statistically significant difference was achieved, we can speculate that a low-volume blood culture sample may have been the cause of this finding. Furthermore, in clinical practice it is not uncommon that blood volumes smaller than ideal are sent for culture from smaller preterm babies, which decreases the diagnostic yield of this important diagnostic tool. Although blood culture is considered by many to be the gold standard in the diagnosis of sepsis, its reported sensitivity ranges from 40% to 86% and depends greatly on sample volume.11,24,25,27 It is clear that patients affected by sepsis often do not have positive cultures. Nevertheless, it is accepted that deciding on purely clinical grounds whether a patient is septic or not is inadequate; thus, we must rely on all available diagnostic tools.12,28 Our inclusion criteria allowed us to exclude potential origins of shock other than sepsis, such as cardiogenic, neurogenic, endocrine, hypovolemic and traumatic shock. Our findings are in agreement with those of Gibot S. et al (2004), who stated that in adults, a plasma sTREM-1 level higher than 60 pg/mL yields a sensitivity of 96%, specificity of 89%, positive likelihood ratio (+LR) of 8.6 and negative likelihood ratio (–LR) of 0.04 for diagnosing patients with systemic inflammatory response and sepsis.4 In a related paper published earlier the same year, their group described a trend of higher sTREM-1 levels in subjects who died than in survivors, and concluded that the detection of sTREM-1 in bronchoalveolar lavage fluid was useful for establishing or excluding the diagnosis of bacterial or fungal pneumonia.3 In a meta-analysis developed to evaluate the accuracy of sTREM-1 as a diagnostic test for bacterial infection, among 73 studies with adult populations, 13 were selected, with a total of 980 patients included in the analysis. The median age was 60 years (IQR = 46 to 67 years), and the prevalence of bacterial infections ranged from 34 to 80% across the studies. The metaanalysis showed that sTREM-1 can be used as a helpful diagnostic criterion for bacterial infection, except for urinary tract infections, in adult populations. The importance of the negative likelihood ratio is well supported, and it further reduces the post-test probability, allowing researchers to rule out a possible infection.6 Zhang J. et al. also found that sTREM-1 levels accurately reflect the severity of sepsis and are a sensitive prognostic tool.7 A definite weakness of our approach was our attempt to establish sensitivity and specificity values using the same data-set from which we obtained the proposed cut-off value of sTREM-1 as a predictor of septic shock. When we derive a cut-off value from an existing data-set, it is expected to perform well as a discriminator in the same data-set.29 To validate our findings, future studies with larger sample sizes and a prospective evaluation of a cut-off must be performed to justify the derivation of important conclusions using this diagnostic tool and to establish the best threshold value of sTREM-1. Sensitivity, specificity and disease prevalence – and therefore +LR and –LR – all interact to determine a test’s ability to predict the presence or absence of any certain condition (disease or outcome).29 It is necessary to continue studying the prognostic potential of sTREM-1 as a predictor of septic shock in our setting. Our next step is to perform an extended study to confirm or disprove our findings and to explore the predictive potential of sTREM-1 in a larger cohort of neonates affected with LONS with and without concomitant septic shock/death. Furthermore, sTREM-1 should be compared to other biomarkers that have been shown to be sensitive for the detection of severe sepsis and septic

31

shock in adults, such as procalcitonin, soluble urokinase-type plasminogen activator receptor and ST2, alone or as part of a panel of biomarkers.30 5. Conclusion In conclusion, sTREM-1 has the potential to provide excellent predictive value. Further trials with larger sample sizes are needed to identify the optimal cut-off value and to establish diagnostic accuracy. Acknowledgments The authors gratefully acknowledge the infants and their parents who participated in this study. We also thank the nurses and attending physicians for their assistance, which allowed us to conduct this study as planned. We wish to recognize the expert and opportune technical assistance of Maria Isabel Garcı´a-Cruz and Gonzalo Gutie´rrezOrozco, certified laboratory professionals, who conducted and supervised the laboratory work. Competing Interests: All authors expressly deny having any competing interests regarding study materials, supplies or techniques. Funding: This work was funded solely by institutional research funds. All authors declare no conflicts of interest. This study was supported solely by institutional research funds from: Tecnolo´gico de Monterrey, Escuela de Medicina, Ca´tedra de Investigacio´n en Crecimiento y Desarrollo Humano. Presented in part, as an oral dissertation, at 51st Latin American Society for Pediatric Research (SLAIP) meeting, in Valparaı´so, Chile, Oct. 23, 2013 References 1. Bouchon A, Dietrich J, Colonna M. Cutting edge: inflammatory responses can be triggered by TREM-1, a novel receptor expressed on neutrophils and monocytes. J Immunol 2000;164(10):4991–5. 2. Bouchon A, Facchetti F, Weigand MA, Colonna M. TREM-1 amplifies inflammation and is a crucial mediator of septic shock. Nature 2001;410(6832):1103–7. 3. Gibot S, Cravoisy A, Levy B, Bene MC, Faure G, Bollaert PE. Soluble triggering receptor expressed on myeloid cells and the diagnosis of pneumonia. N Engl J Med 2004;350(5):451–8. 4. Gibot S, Kolopp-Sarda MN, Be´ne´ MC, Cravoisy A, Levy B, Faure GC, et al. Plasma level of a triggering receptor expressed on myeloid cells-1: its diagnostic accuracy in patients with suspected sepsis. Ann Intern Med 2004;141(1):9–15. 5. Gibot S, Cravoisy A. Soluble form of the triggering receptor expressed on myeloid cells-1 as a marker of microbial infection. Clin Med Res 2004;2(3):181–7. 6. Jiyong J, Tiancha H, Wei C, Huahao S. Diagnostic value of the soluble triggering receptor expressed on myeloid cells-1 in bacterial infection: a meta-analysis. Intensive Care Med 2009;35(4):587–95. 7. Zhang J, She D, Feng D, Jia Y, Xie L. Dynamic changes of serum soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) reflect sepsis severity and can predict prognosis: a prospective study. BMC Infect Dis 2011;11:53. 8. Stoll BJ, Hansen N, Fanaroff AA, Wright LL, Carlo WA, Ehrenkranz RA, et al. Lateonset sepsis in very low birth weight neonates: the experience of the NICHD Neonatal Research Network. Pediatrics 2002;110(2 Pt 1):285–91. 9. Kermorvant-Duchemin E, Laborie S, Rabilloud M, Lapillonne A, Claris O. Outcome and prognostic factors in neonates with septic shock. Pediatr Crit Care Med 2008;9(2):186–91. 10. Sarafidis K, Soubasi-Griva V, Piretzi K, Thomaidou A, Agakidou E, Taparkou A, et al. Diagnostic utility of elevated serum soluble triggering receptor expressed on myeloid cells (sTREM)-1 in infected neonates. Intensive Care Med 2010; 36(5):864–8. 11. Goldstein B, Giroir B, Randolph A. Pediatr Crit Care Med 2005;6(1):2–8. 12. Haque KN. Definitions of bloodstream infection in the newborn. Pediatr Crit Care Med 2005;6(3 Suppl):S45–9. 13. Mahieu LM, De Muynck AO, De Dooy JJ, Laroche SM, Van Acker KJ. Prediction of nosocomial sepsis in neonates by means of a computer-weighted bedside scoring system (NOSEP score). Crit Care Med 2000;28(6):2026–33. 14. Mahieu LM, De Dooy JJ, Cossey VR, Goossens LL, Vrancken SL, Jespers AY, et al. Internal and external validation of the NOSEP prediction score for nosocomial sepsis in neonates. Crit Care Med 2002;30(7):1459–66. ˜ o-Va´zquez R, Ortiz-Ibarra FJ. Validacio´n de la escala 15. Reyna-Figueroa J, Brisen NOSEP-1 para el diagno´stico de sepsis nosocomial en recie´n nacidos prema-

32

16.

17.

18.

19.

20.

21.

22.

V. Arı´zaga-Ballesteros et al. / International Journal of Infectious Diseases 30 (2015) 27–32 turos menores de 1500 g. Bol Med Hosp Infant Mexico 2005;62(septiembre– octubre):321–8. Cohen-Wolkowiez M, Moran C, Benjamin DK, Cotten CM, Clark RH, Smith PB. Early and late onset sepsis in late preterm infants. Pediatr Infect Dis J 2009; 28(12):1052–6. Vergnano S, Menson E, Kennea N, Embleton N, Russell AB, Watts T, et al. Neonatal infections in England: the NeonIN surveillance network. Arch Dis Child Fetal Neonatal Ed 2011;96(1):F9–14. Grisaru-Soen G, Friedman T, Dollberg S, Mishali H, Carmeli Y. Late-onset bloodstream infections in preterm infants: a 2-year survey. Pediatr Int 2012;54(6):748–53. Mazzucchelli I, Garofoli F, Ciardelli L, Borghesi A, Tzialla C, Di Comite A, et al. Diagnostic performance of triggering receptor expressed on myeloid cells-1 and CD64 index as markers of sepsis in preterm newborns. Pediatr Crit Care Med 2013;14(2):178–82. Adly AA, Ismail EA, Andrawes NG, El-Saadany MA. Circulating soluble triggering receptor expressed on myeloid cells-1 (sTREM-1) as diagnostic and prognostic marker in neonatal sepsis. Cytokine 2014;65(2):184–91. Garofoli F, Borghesi A, Mazzucchelli I, Tzialla C, Di Comite A, Tinelli C, et al. Preterm newborns are provided with triggering receptor expressed on myeloid cells-1. Int J Immunopathol Pharmacol 2010;23(4):1297–301. Matoba N, Yu Y, Mestan K, Pearson C, Ortiz K, Porta N, et al. Differential patterns of 27 cord blood immune biomarkers across gestational age. Pediatrics 2009; 123(5):1320–8.

23. Colonna M, Facchetti F. TREM-1 (triggering receptor expressed on myeloid cells): a new player in acute inflammatory responses. J Infect Dis 2003; 187(Suppl 2):S397–401. 24. Lara-Dı´az VJ, De La Vega-Me´ndez J, Arı´zaga-Ballesteros V, Tinoco-Torres BR, Moreno-Cuevas JE. Diagnostic accuracy of buffy coat culture compared to total blood culture in late-onset sepsis of the newborn. Int J Infect Dis 2013;17(2): e110–e114. 25. Schelonka RL, Chai MK, Yoder BA, Hensley D, Brockett RM, Ascher DP. Volume of blood required to detect common neonatal pathogens. J Pediatr 1996;129(2): 275–278. 26. Kellogg JA, Ferrentino FL, Goodstein MH, Liss J, Shapiro SL, Bankert DA. Frequency of low level bacteremia in infants from birth to two months of age. Pediatr Infect Dis J 1997;16(4):381–5. 27. Bouza E, Sousa D, Rodriguez-Creixems M, Lechuz JG, Munoz P. Is the volume of blood cultured still a significant factor in the diagnosis of bloodstream infections? Journal of clinical microbiology 2007;45(9):2765. 28. Fischer JE, Seifarth FG, Baenziger O, Fanconi S, Nadal D. Hindsight judgement on ambiguous episodes of suspected infection in critically ill children: poor consensus amongst experts? Eur J Pediatr 2003;162(12):840–3. 29. Akobeng AK. Understanding diagnostic tests 2: likelihood ratios, pre- and posttest probabilities and their use in clinical practice. Acta Paediatr 2007;96(4): 487–491. 30. Moyer MW. New biomarkers sought for improving sepsis management and care. Nat Med 2012;18(7):999.