Secondary Logo

Journal Logo

Original Articles: Gastroenterology

Detection of Sepsis in Preterm Infants by Fecal Volatile Organic Compounds Analysis: A Proof of Principle Study

Berkhout, Daniel J.C.∗,†; Niemarkt, Hendrik J.; Buijck, Martin; van Weissenbruch, Mirjam M.§; Brinkman, Paul||; Benninga, Marc A.; van Kaam, Anton H.; Kramer, Boris W.#; Andriessen, Peter‡,#; de Boer, Nanne K.H.∗∗; de Meij, Tim G.J.

Author Information
Journal of Pediatric Gastroenterology and Nutrition: September 2017 - Volume 65 - Issue 3 - p e47-e52
doi: 10.1097/MPG.0000000000001471

Abstract

What Is Known?

  • An electronic nose can detect fecal volatile organic compound profiles, which are considered to reflect intestinal microbiota composition and function.
  • Development of late-onset sepsis in preterm infants cannot solely be attributed to intravascular catheters.

What Is New?

  • Fecal volatile organic compound profiles differ significantly between preterm infants with late-onset sepsis and controls before clinical onset of sepsis, indicating that the gut may be involved in at least a selection of late-onset sepsis cases.
  • As the altered volatile organic compound profiles are found before onset of late-onset sepsis, this opens a window of opportunity to prevent sepsis by altering gut microbiota in preventive strategies.

Despite advanced technology and specialized care at neonatal intensive care units, incidence rates of late-onset sepsis (LOS, onset >72 hours after birth) in premature infants has increased over the past decades toward 20% (1,2). LOS is associated with prolonged hospital admission and associated morbidities including patent ductus arteriosus, bronchopulmonary dysplasia, and necrotizing enterocolitis (NEC), and most importantly, impaired neurodevelopmental outcome (3,4). Although overall mortality rates in premature infants have declined over the past years, sepsis-related mortality rates remained unchanged (5,6).

Although intravascular catheters are considered the major source of LOS (7), recent studies on LOS have demonstrated genetic incongruity between organisms isolated from the blood culture and bacteria cultured from the intravascular catheter tip (8,9). The genetic similarity between cultured LOS pathogens and isolates from the gastrointestinal tract points to the origin from the gut (10–16). These isolated pathogens from blood cultures could be detected in the gut already several days before the clinical onset of LOS (10,13,14,16). These findings have led to the hypothesis that in at least a part of the LOS cases, the gut instead of foreign bodies serves as the primary source of infection (10–17). The precise mechanisms are not understood. The translocation from the gut into mesenteric lymph nodes and ultimately into the bloodstream may be mediated by intestinal hypoperfusion leading to a disturbed mucosal barrier function of the already immature gut, low bacterial species diversity supporting overgrowth of potentially invasive microbes, suboptimal nutritional condition, and immaturity of the immune system (18–21).

Therefore, intestinal microbiota analysis may have potential as an early noninvasive diagnostic biomarker for LOS. Because traditional culture techniques are usually insufficient to detect the common LOS pathogens in fecal samples (strains of coagulase-negative Staphylococcus [CoNS], S aureus, Escherichia coli, Klebsiella), novel molecular techniques are needed to unravel the highly complex intestinal microbiota composition. Implementation of such techniques into daily clinical practice is, however, hampered by the scarcity of necessary equipment, inefficiency in time and high costs, and the inability to generate results within acceptable time frames from a clinical point of view. The search for new, preferably noninvasive biomarkers therefore continues.

Fecal volatile organic compounds (VOCs) may offer a solution in this dilemma. VOCs are primarily generated by microbial metabolic processes and host metabolism and are therefore considered to, in part, reflect gut microbiota composition. VOCs are carbon-based chemicals originating from both physiological and pathophysiological metabolic processes in the human body (22). They can be analyzed by either chemical analytical techniques, including gas chromatography-mass spectrometry (GC-MS), enabling identification of individual VOCs, or handheld electronic nose (eNose) devices (23). We hypothesized that LOS is preceded by dysbiosis in the gut and therefore may be detected by fecal VOC analysis. The aim of the present study was to compare VOC profiles obtained from fecal samples collected of preterm infants with LOS in the days before clinical onset of disease, with VOC profiles of matched controls, as measured by eNose technology.

METHODS

Subjects

Preterm infants with a gestational age of <30 weeks and born between September 2013 and June 2015 at the Neonatal Intensive Care Units (NICUs) of the VU Medical Centre and Academic Medical Centre in Amsterdam, and the Máxima Medical Centre in Veldhoven, were eligible to participate in this prospective case-control study. Subjects with congenital intestinal anomalies (eg, anus atresia, Hirschsprung disease) and surgery of the gastrointestinal tract were excluded. In addition, preterm infants who developed NEC (any stage) during the first 28 days of life were excluded from the present study. Standard demographic and clinical data, including mode of delivery, enteral and parenteral feeding pattern, presence of central venous catheter (CVC), medication, respiratory support, and clinical condition (including sepsis and NEC) was prospectively collected. None of the participating NICUs routinely administered probiotics to the studied population. The study was approved by the local institutional review boards of all participating centers, and written informed consent was obtained from all parents of included patients.

Sample Size Calculation

Based on the results of our former studies on fecal VOC analysis, we concluded that per selected time interval a minimal of 10 subjects per study group was required to obtain a power (β-error) of 0.80 to reject the null hypothesis that no differences between the VOC profiles of patients with LOS and matched controls exists, with an P value (α-error) at <0.05 (24–26).

Sample Collection

From birth until the postnatal age of 28 days, fecal samples were collected daily from the diaper by a nurse. When an infant was discharged from the NICU, transferred to another hospital or deceased before the age of 28 days, stool collection was stopped. If a subject passed more than 1 fecal sample per day, only the first produced stool was collected. If bowel movements were absent during a day, the subsequently produced fecal sample was collected. Fecal samples were stored in a stool container (Stuhlgefäß 10 mL, Frickenhausen, Germany) at a temperature of −20°C.

Patient Cohort and Sample Selection

Preterm infants were allocated to the LOS group if the following Vermont Oxford criteria were met: presence of clinical signs of systemic infection (such as apnea, temperature instability, feeding intolerance, worsening respiratory distress, hemodynamic instability), a positive blood culture, which was obtained ≥72 hours after birth, and antibiotic treatment based on the cultured pathogen (27). Controls were defined as healthy preterm neonates without any clinical symptoms or signs of systemic infection for which appropriate antibiotic treatment was administered.

Fecal samples produced from 1 day up to 5 days before the day of clinical onset of LOS (T0) were correspondingly distributed over five 24-hour time intervals (T−1 up to T−5, respectively). Samples produced at T0 were not used for further analysis to minimize the risk of type I error, because it could not reliably be ascertained whether these samples were produced before the start of antibiotics. Each fecal LOS sample was matched to 1 control sample based on: center of birth, gestational age, birth weight, postnatal age, feeding pattern, and number of days exposed to antibiotics before T0.

Volatile Organic Compound Analysis by eNose

Fecal samples were analyzed by means of Cyranose 320 (Smiths Detection, Pasadena, CA). For a detailed description of the sampling and measurement procedure, we refer to our previous study (24). In brief, approximately 0.5 g of frozen feces was put in a sealed vacutainer (BD Vacutainer 3 mL, Franklin Lakes, NJ) and placed in a 37°C stove for 1 hour, allowing VOCs to fill the headspace. Afterwards, 2 needles were placed into the cap of the sealed vacutainer, which were subsequently incorporated into an airtight system attached to the eNose. After a baseline reference was created by connection of a VOC filter (A1, North Safety, Middelburg, The Netherlands), the actual measurement was performed. Fecal VOCs were led across an array of 32 sensors, all comprising conductive carbon black material, which is blended throughout a specific nonconducting polymer (28). These polymer coatings swell when VOCs are presented, subsequently increasing the electrical resistance by augmenting the distance between the conductive carbon black material particles (28). Each individual VOC interacts with multiple sensors and each sensor interacts with different fractions of the VOC mixture. Each polymer coating being unique, eventually leads to 32 unique resistance changes, resulting in a so-called “smellprint” (28). Based on pattern recognition algorithms, these smell prints can be used to differentiate between clinical subgroups.

Data Analysis

Clinical and demographic data were compared by independent t test, nonparametric test, or chi-square test where appropriate. Data analysis was performed using R-script (Rstudio [version 0.99.489] engine by R [version 3.2.2], using R packages: modern applied statistics with S, stats, surrogate variable analysis, package receiver operating characteristic). To minimize variance based on systematic nonbiological differences between analyzed groups (batches), ComBat algorithm was applied before the analysis. This function allows for removal of batch effects and other unwanted sources of variation (29). After batch correction, principal component analysis was used to reduce data from 32 individual sensors into a set of 4 principal components (PCs), capturing the largest variance in the dataset. An independent t test was performed to assess discriminating PCs. These PCs were subsequently used in an internally crossvalidated canonical discriminant analysis based on the leave one out method, to calculate the probability of belonging to the LOS group. An ROC curve with associated 95% confidence interval, P value, sensitivity (%), and specificity (%) was constructed for the test set and after internal validation was performed. A P value of <0.05 was considered statistically significant.

RESULTS

Patient Population

A total of 248 preterm infants were consecutively included during the study period (80 sepsis, 168 subjects without sepsis). Based on the strict matching criteria, we could subsequently select 76 preterm infants for further VOC analysis (36 sepsis cases matched to 40 controls), together providing 156 fecal samples (Table 1). In 6 infants allocated to the LOS group (4 S capitis, 2 S epidermidis) treatment consisted of central line removal followed by 72 to 96 hours of antibiotic treatment, in all other cases antibiotics was administered for a minimum of 5 days. In the LOS cohort, 78% of isolated pathogens were CoNS species (Table 2). Two (6%) of 36 neonates with LOS died within the first 28 days of life and in both cases E coli was isolated from the blood culture. None of the infants assigned to the control group died during the study period. An overview of patient characteristics of the 2 subgroups is depicted in Table 3.

TABLE 1
TABLE 1:
Number of fecal samples in each group per selected time point used for volatile organic compound analysis. Selected time points are 1 (T-1), 2 (T-2), 3 (T-3), 4 (T-4), and 5 (T-5) days before diagnosis
TABLE 2
TABLE 2:
Isolated pathogens (n = [%]) from blood cultures in 36 sepsis patients
TABLE 3
TABLE 3:
Subject characteristics of the 2 subgroups sepsis and controls

Fecal Gas Analysis

Based on their fecal VOC profiles, patients with LOS could statistically significantly be discriminated from strictly matched controls, at T−3 (area under the curve [±95% confidence interval], P value, sensitivity, specificity; 70.2 [52.2–88.3], 0.0329, 57.1%, 61.5%), T−2 (77.7 [62.7–92.7], 0.0496, 75.0%, 70.8%) and T−1 (70.4 [49.6–91.3], 0.0369, 64.3%, 64.3%). This discrimination was not possible for T−4 (60.9 [38.0–83.8], 0.5968, 45.5, 46.7) and T−5 (63.2 [39.8–86.6], 0.3443, 70.0, 64.3). An overview of the performance characteristics for the discrimination of LOS from healthy controls is given in Table 4.

TABLE 4
TABLE 4:
Performance characteristics with corresponding sensitivity and specificity of fecal volatile organic compound analysis for the discrimination of late-onset sepsis and controls

Notably, not all preterm infants with LOS did pass stools on each day before LOS onset. We performed a post-hoc analysis to assess whether fecal VOC profiling could be of value to predict LOS, regardless of the number of days that the last fecal sample was produced/collected before LOS onset. Each LOS infant and matched control with fecal samples collected on either 3, 2, or 1 day before clinical onset was included. Fecal samples closest to T0 were selected for analysis. A total of 32 LOS samples and matched controls were available, the number of samples per time interval used is depicted in Supplemental Digital Content 1, Table 1, http://links.lww.com/MPG/A855. VOC profiles of LOS subjects could statistically significantly be discriminated from controls, irrespective of the day of last stool passage before LOS onset (66.8 [53.2–80.4], 0.0242, 64.5%, 63.6%).

A post-hoc analysis was also performed to assess whether differences in VOC profiles of the 2 groups changed when only CoNS cases were selected. Preterm infants with CoNS sepsis (n = 28) could be discriminated from strictly matched controls at T−1 (70.4 [48.4–92.5], 0.0377, 66.7%, 64.3%), but not at all other preselected time intervals (see Supplemental Digital Content 2, Table 2, http://links.lww.com/MPG/A856).

DISCUSSION

In the present study, we have compared fecal VOC profiles of infants with LOS, up to 5 days before clinical onset, with VOC profiles of matched controls. We found statistically significant differences in fecal VOC profiles between the 2 subgroups, from up to 3 days, but not earlier at 4 and 5 days before clinical onset of LOS.

Over the past decades there has been a rapidly increasing interest in the potential of VOCs as diagnostic biomarker for several diseases, including malignancies, inflammatory, metabolic, and infectious diseases (30–37). Several human and animal studies have evaluated the value of VOC as diagnostic biomarker for sepsis, using blood, exhaled breath, or tracheal aspirate (38–42). In only 1 study, including preterm ventilated infants, samples were collected before onset of sepsis (38). In 2 of 8 subjects with bloodstream infections, tracheal aspirate samples were collected before and shortly after diagnosis and subsequently analyzed by an electronic nose (Cyranose320). Remarkably, VOC profiles from samples obtained before diagnosis clustered with samples from infants without sepsis, whereas samples obtained shortly after diagnosis clustered within the sepsis group. This may implicate that, although number of included subjects was low, VOCs analysis of tracheal aspirates do not have the potential to detect LOS in preclinical stage.

In a recent study from our research group on fecal gas analysis in preterm infants we observed that VOC signatures of infants with NEC (n = 13) could be discriminated from strictly matched controls (n = 14) and from infants with LOS (n = 31), up to 3 days before clinical onset of NEC (26). Study design did not, however, allow for reliable comparison between LOS and controls, because both subgroups could not strictly be matched as groups were primarily matched to NEC cases.

The observed differences in fecal VOC profiles between LOS cases and healthy matched controls in our study are consistent with findings from studies on altered intestinal microbiota composition before LOS onset (10–16). This may be explained by the fact that fecal VOCs are considered to largely reflect microbiota composition and its interaction with the host (43). In these studies, causative pathogens could be isolated from the gut ranging from 10 up to 1 day preceding LOS onset (10,14), with detection rates varying between 54% and 82% (10,13,14,16). Furthermore, microbial diversity in LOS subjects was found to be lower preceding a sepsis episode compared with controls (11,12), next to an acquired predominance for the phyla Firmicutes and Proteobacteria(11–13). In healthy controls, microbial diversity and abundance of obligate anaerobes, including Clostridium, Klebsiella, and Veillonella species, increases over time (11,13).

In the present study, fecal VOC analysis allowed for statistically significant discrimination between controls and LOS in a preclinical phase, but the clinical significance of this outcome may be questioned. Obviously, observed accuracies are currently insufficient to introduce VOC analysis as a diagnostic biomarker for LOS in daily clinical practice. There are several causes for the relatively modest accuracy to discriminate both subgroups.

Firstly, results of our study may be influenced by the origin of the pathogens in LOS; the assumption that the majority of LOS cases are caused by indwelling devices has been debated in several recent studies (8,9). In 30% of blood stream infection, the causative pathogen could not be isolated from the CVC (8,9), whereas another study demonstrated that the route of acquisition could not be determined in 70% of blood stream infection cases (44). Because LOS can probably not solely be attributed to the presence of indwelling devices nor to bacterial translocation from the gut, early diagnostic biomarkers for LOS entirely focusing on the gut, such as fecal VOC analysis in the present study, will lead to suboptimal results.

Secondly, another possibility for the modest accuracy was the relatively small samples size in combination with the high heterogeneity of the study group, reflected by many different isolated pathogens, hampering better classification. Although majority of LOS cases were caused by CoNS species, apparently creating a homogenous population, it has been demonstrated that each CoNS strain produces a unique combination of VOCs (45). Presence of species-specific metabolic fingerprints may, however, offer opportunities to identify causative pathogens based on their VOC profiles (46).

Thirdly, because microbial composition is highly variable within the first months of life (20), differences between subjects in postnatal age at day of diagnosis may augment interindividual differences in VOC profiles within both study groups, which may have contributed to the only modest accuracy. Future studies, including larger number of subjects, are needed to longitudinally assess day-to-day changes in VOC profiles in early life, including influences of environmental factors, such as diet, medication, and variation in microbial colonization on VOC outcome.

Strengths of the present study are the prospective multicenter design and the inclusion of cases and controls matched for center and clinical demographics. The present study has also several limitations. First, the study was designed as a feasibility study for VOC profiling of neonatal sepsis and represents a relatively small sample size. Second, fecal samples were not available from all subjects at every time point, leading to even smaller sample sizes at predefined time points. It would, therefore, be interesting to evaluate whether VOC profiles from rectal swabs could be used instead of fecal samples, because these can be harvested at any time. Future studies, comprising larger number of subjects are needed to externally validate our observations. Because intestinal microbiota colonization and consequently VOC fingerprints are influenced by many different environmental factors, including medication and feeding pattern, this external validation study should preferably be performed in multicenter setting to minimize the risk of type I error (16).

Next to validation of our observations, future studies should aim at identifying LOS-specific VOCs, using chemical analytical techniques, such as GC-MS. Although such techniques are expensive, time-consuming, and require highly trained personnel, limiting its use as screening tool in clinical practice, this step will allow for development of tailor-made eNose devices that can be implemented in daily NICU practice. In contrast to GC-MS, a primed, disease-specific eNose can be applied in clinical practice because of its relatively low costs and high-throughput capacities, enabling real-time and bedside analysis. The present study demonstrates that analysis of fecal VOCs may be able to reflect/detect gut dysbiosis, which is not only present before the development of NEC but also before onset of LOS (26). Therefore, fecal VOC analysis by an eNose holds promise to become a noninvasive, time, and cost-effective method in the NICU to detect gut dysbiosis. Because the altered VOC profiles are found before onset of LOS, this opens a window of opportunity to prevent sepsis by altering gut microbiota by preventive strategies such as probiotics administration or individualized, targeted antibiotic treatment.

In conclusion, we observed in this proof of principle study that fecal VOC analysis by eNose allowed for discrimination between preterm infants with LOS and matched controls, up to 3 days before clinical onset of LOS. These findings underline the hypothesis that, at least in a selection of LOS, intestinal microbiota plays an etiological role in LOS. The potential of fecal VOC profiling in the early detection of LOS needs to be explored in future studies.

ACKNOWLEDGMENTS

The authors would like to thank Heidi Theeuwen, Ilse de Lange, and Mirjam van de Velde for their help in collecting clinical data and fecal samples.

REFERENCES

1. Bizzarro MJ, Raskind C, Baltimore RS, et al. Seventy-five years of neonatal sepsis at Yale: 1928–2003. Pediatrics 2005; 116:595–602.
2. Boghossian NS, Page GP, Bell EF, et al. Late-onset sepsis in very low birth weight infants from singleton and multiple-gestation births. J Pediatr 2013; 162:1120–1124.
3. 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 (2 pt 1):285–291.
4. Stoll BJ, Hansen NI, Adams-Chapman I, et al. Neurodevelopmental and growth impairment among extremely low-birth-weight infants with neonatal infection. JAMA 2004; 292:2357–2365.
5. Berrington JE, Hearn RI, Bythell M, et al. Deaths in preterm infants: changing pathology over 2 decades. J Pediatr 2012; 160:49–53.
6. Patel RM, Kandefer S, Walsh MC, et al. Causes and timing of death in extremely premature infants from 2000 through 2011. N Engl J Med 2015; 372:331–340.
7. Sohn AH, Garrett DO, Sinkowitz-Cochran RL, et al. Prevalence of nosocomial infections in neonatal intensive care unit patients: results from the first national point-prevalence survey. J Pediatr 2001; 139:821–827.
8. Garland JS, Alex CP, Sevallius JM, et al. Cohort study of the pathogenesis and molecular epidemiology of catheter-related bloodstream infection in neonates with peripherally inserted central venous catheters. Infect Control Hosp Epidemiol 2008; 29:243–249.
9. Mueller-Premru M, Gubina M, Kaufmann ME, et al. Use of semi-quantitative and quantitative culture methods and typing for studying the epidemiology of central venous catheter-related infections in neonates on parenteral nutrition. J Med Microbiol 1999; 48:451–460.
10. Carl MA, Ndao IM, Springman AC, et al. Sepsis from the gut: the enteric habitat of bacteria that cause late-onset neonatal bloodstream infections. Clin Infect Dis 2014; 58:1211–1218.
11. Madan JC, Salari RC, Saxena D, et al. Gut microbial colonisation in premature neonates predicts neonatal sepsis. Arch Dis Child Fetal Neonatal Ed 2012; 97:F456–F462.
12. Mai V, Torrazza RM, Ukhanova M, et al. Distortions in development of intestinal microbiota associated with late onset sepsis in preterm infants. PLoS One 2013; 8:e52876.
13. Shaw AG, Sim K, Randell P, et al. Late-onset bloodstream infection and perturbed maturation of the gastrointestinal microbiota in premature infants. PLoS One 2015; 10:e0132923.
14. Soeorg H, Huik K, Parm U, et al. Genetic relatedness of coagulase-negative Staphylococci from gastrointestinal tract and blood of preterm neonates with late-onset sepsis. Pediatr Infect Dis J 2013; 32:389–393.
15. Stewart CJ, Marrs EC, Magorrian S, et al. The preterm gut microbiota: changes associated with necrotizing enterocolitis and infection. Acta Paediatr 2012; 101:1121–1127.
16. Taft DH, Ambalavanan N, Schibler KR, et al. Center variation in intestinal microbiota prior to late-onset sepsis in preterm infants. PLoS One 2015; 10:e0130604.
17. Sertaridou E, Papaioannou V, Kolios G, et al. Gut failure in critical care: old school versus new school. Ann Gastroenterol 2015; 28:309–322.
18. Sherman MP. New concepts of microbial translocation in the neonatal intestine: mechanisms and prevention. Clin Perinatol 2010; 37:565–579.
19. Wiest R, Rath HC. Gastrointestinal disorders of the critically ill. Bacterial translocation in the gut. Best Pract Res Clin Gastroenterol 2003; 17:397–425.
20. Unger S, Stintzi A, Shah P, et al. Gut microbiota of the very-low-birth-weight infant. Pediatr Res 2015; 77:205–213.
21. Marchant EA, Boyce GK, Sadarangani M, et al. Neonatal sepsis due to coagulase-negative staphylococci. Clin Dev Immunol 2013; 2013:586076.
22. Van der Schee MP, Paff T, Brinkman P, et al. Breathomics in lung disease. Chest 2015; 147:224–231.
23. Buijck M, Berkhout DJ, De Groot EF, et al. Sniffing out paediatric gastro-intestinal diseases: the potential of volatile organic compounds as biomarkers for disease. J Pediatr Gastroenterol Nutr 2016; 63:585–591.
24. De Boer NK, De Meij TG, Oort FA, et al. The scent of colorectal cancer: detection by volatile organic compound analysis. Clin Gastroenterol Hepatol 2014; 12:1085–1089.
25. De Meij TG, De Boer NK, Benninga MA, et al. Faecal gas analysis by electronic nose as novel, noninvasive method for assessment of active and quiescent paediatric inflammatory bowel disease: proof of principle study. J Crohns Colitis (2014 Sep 22) pii: S1873-9946(14)00285-2.
26. De Meij TG, Van der Schee MP, Berkhout DJ, et al. Early detection of necrotizing enterocolitis by fecal volatile organic compounds analysis. J Pediatr 2015; 167:562–567.
27. Vermont Oxford Network Database; Manual of Operations: Part 2 Data Definitions and Infant Data Forms. [https://public.vtoxford.org/]. Burlington; Updated 2015 November. https://public.vtoxford.org/wp-content/uploads/2015/09/Manual_of_Operations_Part2_v20.pdf. Accessed May 6, 2016.
28. Li J. The Cyranose chemical vapor analyzer. Sensors 2000; 17:56–61.
29. Leek JT, Johnson WE, Parker HS, et al. The SVA package for removing batch effects and other unwanted variation in high-throughput experiments. Bioinformatics 2012; 28:882–883.
30. Arasaradnam RP, Covington JA, Harmston C, et al. Review article: next generation diagnostic modalities in gastroenterology—gas phase volatile compound biomarker detection. Aliment Pharmacol Ther 2014; 39:780–789.
31. Krilaviciute A, Heiss JA, Leja M, et al. Detection of cancer through exhaled breath: a systematic review. Oncotarget 2015; 6:38643–38657.
32. Gasparri R, Santonico M, Valentini C, et al. Volatile signature for the early diagnosis of lung cancer. J Breath Res 2016; 10:016007.
33. Amal H, Leja M, Funka K, et al. Breath testing as potential colorectal cancer screening tool. Int J Cancer 2016; 138:229–236.
34. Walton C, Patel M, Pitts D, et al. The use of a portable breath analysis device in monitoring type 1 diabetes patients in a hypoglycaemic clamp: validation with SIFT-MS data. J Breath Res 2014; 8:037108.
35. Bond A, Vernon A, Reade S, et al. Investigation of volatile organic compounds emitted from faeces for the diagnosis of Giardiasis. J Gastrointestin Liver Dis 2015; 24:281–286.
36. Arasaradnam RP, McFarlane M, Ling K, et al. Breathomics-exhaled volatile organic compound analysis to detect hepatic encephalopathy: a pilot study. J Breath Res 2016; 10:016012.
37. Cho YS, Jung SC, Oh S. Diagnosis of bovine tuberculosis using a metal oxide-based electronic nose. Lett Appl Microbiol 2015; 60:513–516.
38. Rogosch T, Herrmann N, Maier RF, et al. Detection of bloodstream infections and prediction of bronchopulmonary dysplasia in preterm neonates with an electronic nose. J Pediatr 2014; 165:622–624.
39. Fink T, Wolf A, Maurer F, et al. Volatile organic compounds during inflammation and sepsis in rats: a potential breath test using ion-mobility spectrometry. Anesthesiology 2015; 122:117–126.
40. Guaman AV, Carreras A, Calvo D, et al. Rapid detection of sepsis in rats through volatile organic compounds in breath. J Chromatogr B Analyt Technol Biomed Life Sci 2012; 881–882:76–82.
41. Lim SH, Mix S, Xu Z, et al. Colorimetric sensor array allows fast detection and simultaneous identification of sepsis-causing bacteria in spiked blood culture. J Clin Microbiol 2014; 52:592–598.
42. Edson RS, Rosenblatt JE, Washington JA, et al. Gas-liquid chromatography of positive blood cultures for rapid presumptive diagnosis of anaerobic bacteremia. J Clin Microbiol 1982; 15:1059–1061.
43. Garner CE, Smith S, De Lacy CB, et al. Volatile organic compounds from feces and their potential for diagnosis of gastrointestinal disease. FASEB J 2007; 21:1675–1688.
44. De Brito CS, Von Dolinger de BD, Steffen Abdallah VO, et al. Are there any differences among pathogenesis of catheter-related primary bloodstream infection in adults and neonates patients? Am J Infect Control 2010; 38:494–495.
45. Stavropoulou DA, Borremans W, De VL, et al. Amino acid conversions by coagulase-negative staphylococci in a rich medium: assessment of inter- and intraspecies heterogeneity. Int J Food Microbiol 2015; 212:34–40.
46. Bos LD, Sterk PJ, Schultz MJ. Volatile metabolites of pathogens: a systematic review. PLoS Pathog 2013; 9:e1003311.
Keywords:

electronic nose; gut dysbiosis; late-onset sepsis; microbiota; preterm; volatile organic compound

Supplemental Digital Content

Copyright © 2016 by European Society for Pediatric Gastroenterology, Hepatology, and Nutrition and North American Society for Pediatric Gastroenterology, Hepatology, and Nutrition