Central venous catheter–associated bloodstream infections (CVC-BSIs) are a major cause of morbidity and mortality in the pediatric intestinal failure (IF) population. Although the reported incidence of CVC-BSIs appears to vary across centers (1–7), the presentation of a febrile child with IF and a central catheter generally leads to admission and treatment with broad-spectrum antibiotics. CVC-BSIs further complicate care both directly (8) and indirectly because of their effect on the development of parenteral nutrition–associated liver disease (9). The care of a patient with IF with a CVC is evolving through the use of new technologies in the form of antimicrobial line locks (10–18), coated catheters and dressings (19–22), and selective gut decontamination (23), which are designed to reduce the incidence of CVC-BSIs. Each therapy carries with it potential adverse effects such as catheter occlusion, the development of resistant organisms, and expense. The identification of a serological biomarker that predicts susceptibility to infection would greatly enhance care in this population, allowing the judicious use of these therapies for those patients who are at increased risk for CVC-BSIs.
Several candidate biomarkers have emerged with the potential to identify patients with bacteremia. Two candidates have shown particular promise for the population with IF. The soluble form of the triggering receptor expressed on myeloid cells-1 (sTREM-1), a receptor on the surface of neutrophils and monocytes, is released in the presence of microbial components. sTREM-1 expression is elevated in bacterial illness but low in noninfectious inflammatory disorders (24–28). sTREM-1 has been used to distinguish acute bacterial infections from other inflammatory conditions such as ventilator-assisted pneumonia (26,28,29), bacterial meningitis (30,31), sepsis in adults (26,27,32), and spontaneous bacteremia and urinary tract infection in infancy (33). The production of this marker, independent of hepatic synthetic function, would be expected to increase in the face of a bacterial BSI. Alternatively, lipopolysaccharide (LPS)-binding protein (LBP) is hepatic derived. Although it has generally been associated with Gram-negative infection, it has also been shown to be elevated in Gram-positive and fungal infections (34–37). Made in response to bacterial-derived LPS, LBP forms complexes with LPS or LPS-containing particles in the bloodstream and leads to monocyte and macrophage activation. In adult patients with cirrhosis and ascites, elevated serum LBP is the only independent predictor of spontaneous bacterial peritonitis, thereby suggesting that it is able to be synthesized even in the presence of severe liver disease (38). LBP also reflects long-term exposure to LPS and has been used as a marker for chronic LPS exposure (39). In inflammatory bowel disease, high LPS levels in patients with active disease correlate with elevated LBP in the serum, which normalizes after medical treatment (40). The ease of detection of LBP as opposed to the direct measurement of LPS, its production in the face of severe liver disease, and its link to chronic exposure to bacterial by-products make it a viable candidate to assess CVC-BSIs in the IF population.
These 2 biomarkers have not been used to determine the presence of infection in this population. Before examining their utility in the prediction of the susceptibility of CVC-BSIs in IF, we sought to validate changes during the phase of acute infection. We hypothesized that both sTREM-1 and LBP levels would rise from baseline during an acute infection and would decline upon antibiotic treatment. As a marker of chronic LPS exposure, we further surmised that baseline LBP would be higher in patients with IF than in healthy controls.
PATIENTS AND METHODS
Patient Selection and Experimental Design
Patients were identified using the clinical IF registry at Cincinnati Children's Hospital Medical Center (CCHMC). For this registry population, IF is defined as patients with a primary gastrointestinal disease requiring total parenteral nutrition (TPN) to maintain adequate nutrition, hydration, electrolyte balance, and growth for at least 30 days. Inclusion criteria for the study population were age between 3 months and 4 years, the presence of a CVC, and at least partial PN dependence for 90 consecutive days. Standard Intralipid was used for all PN. The age criteria were used to capture a homogeneous group of patients in terms of underlying diagnosis as well as maturity of the innate immune system. Patients were excluded if they had undergone small bowel, liver/small bowel, or multivisceral transplant, if they had a known underlying immune disorder, a current diagnosis of infection other than CVC-BSI, or if they had been taking immune suppressant medications or systemic antibiotics for more than 24 hours before inclusion. Patients on selective gut decontamination regimens were not excluded. Recruitment occurred prospectively from September 2008 through October 2009 in the CCHMC Comprehensive Nutrition Center, which functions as an intestinal rehabilitation clinic, and during inpatient hospital admissions throughout the study period. Age-matched subjects from the general gastroenterology population, undergoing routine endoscopy for indications most commonly of abdominal pain and vomiting, were recruited for the control population. Control subjects were excluded for clinical evidence of infection including fever; respiratory symptoms; positive blood, urine, or stool culture; the presence of a CVC (to maintain a homogeneous and more true control population); confirmed systemic illness including immune deficiency; systemic antibiotic use at the time of evaluation; and evidence of organic gastrointestinal disease/inflammation based upon abnormal endoscopic or histologic findings.
Once patients were identified and informed consent was obtained, plasma was collected and categorized by patient type/condition into 4 separate groups: IF baseline (at least 60 days following a CVC-BSI), IF with blood culture obtained for fever and suspected sepsis, posttreatment (within 10 days of completing therapy for a CVC-BSI), and non-IF controls. The IF group with blood cultures drawn was further divided into patients with a laboratory-confirmed CVC-BSI (as defined by the National Nosocomial Infections Surveillance System) (41) and those without. Infections with fungal organisms were excluded because of limited data available on expected sTREM-1 values in this setting. Because the study was designed to capture distinct CVC-BSI episodes, multiple samples during distinct CVC-BSIs could be captured from the same individual patient. Although it was desirable to capture baseline, infected, and posttreatment samples longitudinally, samples were collected as they became available.
Blood samples (3 mL collected in ethylenediaminetetraacetic acid tubes) were obtained for patients with IF by accessing the CVC in the clinic or on the inpatient unit. For febrile/infected IF patients, blood cultures were obtained before the initiation of antibiotics. Samples for the sTREM-1 and LBP assays were obtained within 24 hours of initiation of antibiotics for fever. A standard protocol of empiric vancomycin and piperacillin/tazobactam was used for nearly all of the patients. Therapy for all of the infections was not necessarily uniform because it was adjusted based on clinical factors as necessary. For the control patients, blood was collected by venipuncture in the operating room at the time of endoscopy. Plasma was separated and stored at −80°C until the time of analysis. Enzyme-linked immunosorbent assay (ELISA) was performed per the manufacturer's instructions for sTREM-1 using the Human TREM-1 Quantikine Immunoassay (R&D Systems, Minneapolis, MN) (26,29,33,42) and for LBP using the Human LBP ELISA kit (Hycult Biotechnology, Cell Sciences, Canton, MA). For both sTREM-1 and LBP, standard curves were determined using control reagents, and experimental plasma levels performed in duplicate were quantified based upon standard curves. The sTREM-1 assay can be run in approximately 6 hours and the LBP assay can be run in less than 4 hours. Plasma from the IF baseline samples was submitted to the General Clinical Research Center Laboratory at CCHMC for C-reactive protein immunoassay, which has been standardized to internal controls.
Demographic information, including age, sex, primary diagnosis, dates of surgery, length of remaining bowel, presence of retained ileocecal valve, transplant status, inpatient or outpatient status, percentage of energy from PN, and use of gut decontamination therapy (on cycle/off cycle) was also collected. Baseline laboratory data including conjugated bilirubin, platelet count, white blood cell count, albumin, and prealbumin level were collected from the patients’ medical records.
Comparisons of continuous variables (eg, LBP, sTREM-1) among different study phases and between control subjects and patients were performed with a repeated-measures analysis of variance. There were up to 4 phases observed within patients (patient baseline, patient with fever, patient with infection, patient posttreatment), although not all of the phases were observed for all of the patients. Only a single measurement (control) was performed for control subjects. Planned t-contrasts among these 5 phases were performed to assess particular mean differences of interest. These analyses were performed with the MIXED procedure in SAS version 9.2 (SAS Institute, Cary, NC). Linear regression was used to assess correlation between pairs of continuous variables.
Associations between binary outcomes (ie, listed for transplant, transplanted, death) and continuous or binary predictors were assessed with logistic regression. To better illustrate the tradeoff between the sensitivity and specificity of predicting a given binary outcome based on different possible threshold values of a continuous variable, we constructed receiver operating characteristic (ROC) curves and reported the area under these curves. These analyses were performed with the LOGISTIC procedure in SAS version 9.2. Unless otherwise stated, all of the statistical tests and confidence intervals were 2-sided, with α = 0.05.
The present study protocol was approved by the CCHMC institutional review board before enrollment of patients and collection of samples.
Samples were analyzed from 24 pediatric IF patients (mean age 18.2 months, 32% girls) and 11 control patients (mean age 20.5 months, 55% girls). Table 1 provides clinical characteristics of the patients with IF. Samples were divided into 5 categories: control samples (n = 11), baseline IF samples (n = 22), febrile IF samples without BSI (n = 10), IF samples with laboratory-confirmed BSI (n = 17), and IF posttreatment samples (n = 16). Of the 24 patients with IF in the study, 8 had been listed for liver or multivisceral transplant, 4 were transplanted during the course of the study (excluded from further analysis after transplant), and 4 were nontransplant-associated deaths. The 17 BSIs were captured in 10 patients with IF. Table 2 demonstrates clinical information such as height of fever and related symptoms of the 10 culture-negative febrile events. Table 3 demonstrates this clinical information for the 17 infections as well as the 1 fungal infection that was excluded. Nine infections (52.9%) were Gram-negative, 4 (23.5%) were Gram-positive, and 4 (23.5%) were mixed. Twelve infections (70.6%) were enteric organisms. The distribution of infections during the study period was similar to cumulative registry data collected beyond our enrollment period (data not shown). Seven of the 17 infections (in 4 individuals) occurred in patients concomitantly receiving selective gut decontamination of various regimens. The regimens observed in our population were metronidazole alone, nitazoxanide alone, metronidazole with nystatin, and nystatin alone. Five patients had more than 1 infection captured during the study and 1 of these patients had 4 infections. None of the patients had a change in their CVC in between infections and the samples were drawn from the same line. Four of the 7 repeat infections occurred with at least 1 of the organisms being the same as the previous infection. Four of the 5 patients subsequently had their central lines replaced after repeat infection with the same organism.
sTREM-1 and LBP Levels During Culture-positive Infection Events
We first determined the ability of these candidate biomarkers to signal a culture-positive febrile event. Although both candidates have independently been shown to increase during numerous types of infections, this has not been measured in the context of either IF or CVC infections. During a culture-confirmed infection, mean sTREM-1 level (picograms per milliliter) rose over baseline (115.0 ± 51.2 vs 85.9 ± 27.6, P = 0.011) and declined to preinfection levels (115.0 ± 51.2 vs 77.9 ± 29.8, P = 0.003) in the posttreatment arm of the study (Fig. 1A). Interestingly, sTREM-1 levels were higher in Gram-negative (135.3 ± 19.2) and polymicrobial (117.0 ± 13.2) infections than in Gram-positive (67.3 ± 6.1, P = 0.04 and P = 0.01, respectively) infections, in which levels were similar to those of the baseline group (Fig. 1B).
Similarly, mean LBP (micrograms per milliliter) increased during culture-confirmed infections over baseline (79.8 ± 45.4 vs 20.5 ± 11.3, P < 0.001), returning to near baseline at the completion of antibiotic therapy (79.8 ± 45.4 vs 26.2 ± 10.8, P < 0.001) (Fig. 2A). Posttreatment levels were not statistically different from baseline levels. Mean LBP increased significantly over baseline irrespective of the classification of infection (Gram-negative 92.9 ± 17.3, Gram-positive 63.8 ± 19.9, and mixed 66.1 ± 16.6, all P < 0.0001), although LBP levels were slightly lower in the Gram-positive and polymicrobial groups (Fig. 2B). One polymicrobial infection was excluded because of the presence of a fungal organism in addition to enteric bacteria. The sTREM-1 and LBP levels of this infection were within 1 standard deviation of the mean values of the studied infections. There were no exclusively fungal bloodstream infections observed during the study period.
Because all of the results shown thus far were cross-sectional composites based upon an experimental group and not necessarily obtained in longitudinal fashion, there was a concern that data obtained as such may not clearly capture individual responses. To address this issue, we isolated 6 individual subjects with measurements obtained in chronological order (baseline, infection, posttreatment) (Fig. 3). In all of the cases, LBP rose in response to infection and declined to baseline levels. With sTREM-1, this effect was less pronounced and in 1 case, levels actually fell during the acute infection.
Using the composite data, the magnitude of the increase was much greater in LBP (4-fold) versus sTREM-1 (1.3-fold). The response to Gram-positive infections also appeared to be greater with LBP. Finally, chronological measurements showed a propensity for LBP to mark acute infections. These data, in total, placed LBP as a particular candidate of interest as a marker for BSIs in the IF population.
sTREM-1 and LBP Levels During Febrile, Culture-negative Events
For a marker to have useful specificity in the clinical setting, it would be important for it to remain negative during febrile episodes in which blood cultures remain negative. These events were less common than culture-positive episodes. Mean sTREM-1 level (picograms per milliliter) during febrile episodes (85.9 ± 27.6) fell between baseline (78.6 ± 16.9) and infected (102.0 ± 27.6) levels, although it was not significantly different from either (P = 0.223 and P = 0.339, respectively) (Fig. 4A). Mean LBP level (micrograms per milliliter) during febrile, culture-negative episodes (37.6 ± 23.0) also was intermediate between baseline (20.5 ± 11.3) and infected levels (79.8 ± 45.4). As opposed to sTREM-1, however, LBP during febrile episodes was statistically lower than that from subjects with bacteremia (P < 0.001). Febrile LBP levels tended to be higher than baseline measurements, but the magnitude of difference did not reach statistical significance (P = 0.07) because the sample size was small (Fig. 4B).
Based upon these results, ROC curves were generated for sTREM-1 and LBP. The area under the curve for sTREM-1 was 0.57 and 0.82 for LBP. Neither sTREM-1 nor LBP appeared to be adequate predictors of infection. Although this could be reflective of the small sample size, the overlap between “febrile culture-negative” and “febrile culture-positive” levels made it unlikely that these markers would have sufficient specificity for culture-positive events.
sTREM-1 and LBP at Baseline Compared With Non-IF Controls
One potential explanation of the overlap of culture-positive and culture-negative data would be that samples were obtained at different points within each individual's disease course. LBP levels could be affected by chronic exposure to LPS, thus elevating baseline levels, and potentially dampening levels during infection. LBP is believed to mark chronic endotoxin exposure in other disease processes (35,38–40). To assess whether this phenomenon occurred in our subjects, we were careful to compare our baseline samples with samples obtained from a control population without CVCs or IF and unlikely to experience chronic endotoxin exposure.
Mean baseline sTREM-1 (picograms per milliliter) of the patients with IF was not different from that of the control population (85.9 ± 27.6 vs 78.6 ± 16.9, P = 0.56). Alternately, mean baseline LBP (micrograms per milliliter) was significantly higher in the IF population than in the controls (20.5 ± 11.3 vs 12.1 ± 5.8, P = 0.029). If one divided the patients with IF by inpatient or outpatient status at the time of their baseline lab draw, however, the outpatients had the same mean baseline LBP as the control population (13.5 ± 9.2 vs 12.1 ± 5.8, P = 0.69). Conversely, patients with IF admitted to the hospital for nonfebrile reasons such as dehydration, feeding intolerance, or transplant evaluation had a higher mean baseline LBP (27.5 ± 8.7) than both the control population and the outpatients with IF (P < 0.001 and P = 0.002, respectively). LBP, an acute phase protein, did not correlate (r = 0.282) with C-reactive protein in patients with IF at baseline.
Patients with IF who had a history of previous line infections had higher mean baseline LBP (23.5 ± 10.4) than patients with IF who had never had a BSI (10.1 ± 8.3, P = 0.016), likely reflecting a greater history of endotoxin exposure. Additionally, the 8 patients with IF listed for liver or multivisceral transplant had higher mean LBP at baseline (29.6 ± 9.8) than the 14 not listed for transplant (16.2 ± 9.5, P = 0.033).
Baseline LBP was not affected by demographic factors such as sex or age in months, and did not correlate with length of remaining bowel, whether the patient developed an infection during the course of the study, or received a transplant during the course of the study. Furthermore, it did not correlate with biochemical markers such as conjugated bilirubin, platelet count, white blood cell count, albumin, or prealbumin; however, LBP did correlate with the percentage of total energy obtained from TPN (r = 0.57).
CVC-BSIs are a common and important morbidity that occur during the course of therapy for IF. For our primary objective in this pilot study, we found that both sTREM-1 and LBP rise with acute BSI in pediatric IF patients and return to near baseline levels following antibiotic therapy, mirroring previous studies of sTREM-1 and LBP. The relative increase in LBP compared with baseline samples and the increase independent of the type of pathogen isolated suggest that as a marker of acute infection, LBP outperforms sTREM-1 in this population; however, when used to differentiate a culture-negative febrile episode from a confirmed BSI, which is the clinically relevant question, both biomarkers fell short. One potential explanation for this performance was a small sample size between groups (n = 10 and 17, respectively). Although increasing numbers of samples in ROC curves increase the precision, the large variability of values in each category (febrile or infected) likely limited the utility of these markers to differentiate between these 2 clinical groups. Therefore, it is unlikely that increasing the number of subjects or samples studied would demonstrate a discriminative ability for these biomarkers in this population.
A second objective of the study was to investigate the differences between baseline LBP in the IF population and LBP in healthy controls. The mean baseline LBP level was higher in hospitalized IF patients compared with outpatients and controls, suggesting increased endotoxin exposure, potentially via bacterial translocation without culturable bacteremia, in the hospitalized IF group. To investigate LBP in the IF population further, we examined associations of mean baseline LBP to clinical factors. LBP was higher in patients with a history of BSI, in those requiring more TPN, and in those listed for liver or multivisceral transplantation, suggesting that patients with more severe liver disease have had higher chronic endotoxin exposure. This corresponds to the findings of Sondheimer et al (9), who associated frequent and early infections with indirect effect on mortality by advancing the development of liver disease. Together, these findings suggest that LBP may be a useful biochemical marker of clinical disease severity in the pediatric IF population.
Our study has a number of limitations. The sample size was small; however, as stated, increasing the number of subjects or samples may not overcome the variability of the biomarkers. Additionally, results were cross-sectional composites based upon experimental group and not necessarily obtained in longitudinal fashion. Sequential data on several individual subjects show changes in sTREM-1 and LBP as expected with the composite results. Because LBP is both an acute phase reactant that rises with infection and a marker of chronic endotoxin exposure, its elevation may be affected by either process. In future studies, it would be beneficial to obtain several baseline samples for LBP to assess its reliability and reproducibility in each patient with IF for marking chronic exposure rather than subacute events.
In summary, our findings show that increases in both sTREM-1 and LBP occur during infection; however, neither appears to be clinically useful with regard to prediction of bacteremia. Additionally, elevated LBP at baseline appeared to be associated with clinical surrogates of disease severity such as TPN requirement, inpatient status, history of previous BSI, and listing for transplantation. Further evaluation of baseline LBP in a larger population of patients with IF and on multiple occasions during their clinical course is warranted to confirm these findings and determine whether LBP could prospectively predict disease severity and perhaps even the risk of death or transplantation in this population.
1. Diamanti A, Basso MS, Castro M, et al. Prevalence of life-threatening complications in pediatric patients affected by intestinal failure. Transplant Proc
2. Knafelz D, Gambarara M, Diamanti A, et al. Complications of home parenteral nutrition in a large pediatric series. Transplant Proc
3. Moukarzel AA, Haddad I, Ament ME, et al. 230 patient years of experience with home long-term parenteral nutrition in childhood: natural history and life of central venous catheters. J Pediatr Surg
4. Piedra PA, Dryja DM, LaScolea LJ Jr. Incidence of catheter-associated gram-negative bacteremia in children with short bowel syndrome. J Clin Microbiol
5. Terra RM, Plopper C, Waitzberg DL, et al. Remaining small bowel length: association with catheter sepsis in patients receiving home total parenteral nutrition: evidence of bacterial translocation. World J Surg
6. Wales PW, de Silva N, Kim JH, et al. Neonatal short bowel syndrome: a cohort study. J Pediatr Surg
7. Weber TR. Enteral feeding increases sepsis in infants with short bowel syndrome. J Pediatr Surg
8. Han YY, Carcillo JA, Dragotta MA, et al. Early reversal of pediatric-neonatal septic shock by community physicians is associated with improved outcome. Pediatrics
9. Sondheimer JM, Asturias E, Cadnapaphornchai M. Infection and cholestasis in neonates with intestinal resection and long-term parenteral nutrition. J Pediatr Gastroenterol Nutr
10. Broom J, Woods M, Allworth A, et al. Ethanol lock therapy to treat tunnelled central venous catheter-associated blood stream infections: results from a prospective trial. Scand J Infect Dis
11. Cuntz D, Michaud L, Guimber D, et al. Local antibiotic lock for the treatment of infections related to central catheters in parenteral nutrition in children. JPEN J Parenter Enteral Nutr
12. del Pozo JL. Role of antibiotic lock therapy for the treatment of catheter-related bloodstream infections. Int J Artif Organs
13. Fernandez-Hidalgo N, Almirante B, Calleja R, et al. Antibiotic-lock therapy for long-term intravascular catheter-related bacteraemia: results of an open, non-comparative study. J Antimicrob Chemother
14. Fortun J, Grill F, Martin-Davila P, et al. Treatment of long-term intravascular catheter-related bacteraemia with antibiotic-lock therapy. J Antimicrob Chemother
15. Mouw E, Chessman K, Lesher A, et al. Use of an ethanol lock to prevent catheter-related infections in children with short bowel syndrome. J Pediatr Surg
16. Onder AM, Kato T, Simon N, et al. Prevention of catheter-related bacteremia in pediatric intestinal transplantation/short gut syndrome children with long-term central venous catheters. Pediatr Transplant
17. Onland W, Shin CE, Fustar S, et al. Ethanol-lock technique for persistent bacteremia of long-term intravascular devices in pediatric patients. Arch Pediatr Adolesc Med
18. Rijnders BJ, Van Wijngaerden E, Vandecasteele SJ, et al. Treatment of long-term intravascular catheter-related bacteraemia with antibiotic lock: randomized, placebo-controlled trial. J Antimicrob Chemother
19. Borschel DM, Chenoweth CE, Kaufman SR, et al. Are antiseptic-coated central venous catheters effective in a real-world setting? Am J Infect Control
20. Garland JS, Alex CP, Mueller CD, et al. A randomized trial comparing povidone-iodine to a chlorhexidine gluconate-impregnated dressing for prevention of central venous catheter infections in neonates. Pediatrics
21. Ho KM, Litton E. Use of chlorhexidine-impregnated dressing to prevent vascular and epidural catheter colonization and infection: a meta-analysis. J Antimicrob Chemother
22. Niel-Weise BS, Stijnen T, van den Broek PJ. Anti-infective-treated central venous catheters: a systematic review of randomized controlled trials. Intensive Care Med
23. Stoutenbeek CP, van Saene HK, Little RA, et al. The effect of selective decontamination of the digestive tract on mortality in multiple trauma patients: a multicenter randomized controlled trial. Intensive Care Med
24. Bouchon A, Facchetti F, Weigand MA, et al. TREM-1 amplifies inflammation and is a crucial mediator of septic shock. Nature
25. Gibot S, Cravoisy A. Soluble form of the triggering receptor expressed on myeloid cells-1 as a marker of microbial infection. Clin Med Res
26. Gibot S, Cravoisy A, Levy B, et al. Soluble triggering receptor expressed on myeloid cells and the diagnosis of pneumonia. N Engl J Med
27. Gibot S, Kolopp-Sarda MN, Bene MC, et al. Plasma level of a triggering receptor expressed on myeloid cells-1: its diagnostic accuracy in patients with suspected sepsis. Ann Intern Med
28. Huh JW, Lim CM, Koh Y, et al. Diagnostic utility of the soluble triggering receptor expressed on myeloid cells-1 in bronchoalveolar lavage fluid from patients with bilateral lung infiltrates. Crit Care
29. Phua J, Koay ES, Zhang D, et al. Soluble triggering receptor expressed on myeloid cells-1 in acute respiratory infections. Eur Respir J
30. Bishara J, Hadari N, Shalita-Chesner M, et al. Soluble triggering receptor expressed on myeloid cells-1 for distinguishing bacterial from aseptic meningitis in adults. Eur J Clin Microbiol Infect Dis
31. Determann RM, Weisfelt M, de Gans J, et al. Soluble triggering receptor expressed on myeloid cells 1: a biomarker for bacterial meningitis. Intensive Care Med
32. Gibot S, Cravoisy A, Kolopp-Sarda MN, et al. Time-course of sTREM (soluble triggering receptor expressed on myeloid cells)-1, procalcitonin, and C-reactive protein plasma concentrations during sepsis. Crit Care Med
33. Chen HL, Hung CH, Tseng HI, et al. Soluble form of triggering receptor expressed on myeloid cells-1 (sTREM-1) as a diagnostic marker of serious bacterial infection in febrile infants less than three months of age. Jpn J Infect Dis
34. Oude Nijhuis CS, Vellenga E, Daenen SM, et al. Lipopolysaccharide-binding protein: a possible diagnostic marker for Gram-negative bacteremia in neutropenic cancer patients. Intensive Care Med
35. Albillos A, de la Hera A, Gonzalez M, et al. Increased lipopolysaccharide binding protein in cirrhotic patients with marked immune and hemodynamic derangement. Hepatology
36. Ubenauf KM, Krueger M, Henneke P, et al. Lipopolysaccharide binding protein is a potential marker for invasive bacterial infections in children. Pediatr Infect Dis J
37. Blairon L, Wittebole X, Laterre PF. Lipopolysaccharide-binding protein serum levels in patients with severe sepsis due to gram-positive and fungal infections. J Infect Dis
38. Albillos A, de-la-Hera A, Alvarez-Mon M. Serum lipopolysaccharide-binding protein prediction of severe bacterial infection in cirrhotic patients with ascites. Lancet
39. Lequier LL, Nikaidoh H, Leonard SR, et al. Preoperative and postoperative endotoxemia in children with congenital heart disease. Chest
40. Pastor Rojo O, Lopez San Roman A, Albeniz Arbizu E, et al. Serum lipopolysaccharide-binding protein in endotoxemic patients with inflammatory bowel disease. Inflamm Bowel Dis
41. Mayhall C, ed. Surveillance of Nosocomial Infections
. 3rd ed. Philadelphia: Lippincott Williams & Wilkins; 2004: 1659–702.
42. Tzivras M, Koussoulas V, Giamarellos-Bourboulis EJ, et al. Role of soluble triggering receptor expressed on myeloid cells in inflammatory bowel disease. World J Gastroenterol