In industrialized countries, Campylobacter is the most commonly reported bacterial cause of acute gastroenteritis (AGE). Campylobacter jejuni and Campylobacter coli are the most frequent Campylobacter associated with AGE in humans.
Over the past decades, case reports have identified Campylobacter concisus as a cause of AGE. However, in the past 10 years, of the 4 published case-control studies,1–4 only 2 identified a significant association between C. concisus and diarrhea.1,3 And 1 study reported an association between C. concisus and the absence of diarrhea.2 In each of those studies, diagnostic methods and recruitment of cases and controls were different. Moreover, in 1 study3 where an association between C. concisus and diarrhea was found, the age distributions differed among cases and controls (3 times fewer children <5 years in controls than in cases), which is problematic as C. concisus is more frequently isolated from young children.1
C. concisus is genetically highly diverse. A variety of genotyping methods have been used and help differentiate between pathogenic and nonpathogenic C. concisus; yet, genotyping methods have not been able to provide perfect separation of cases and controls.5
The present study, therefore, had 2 aims. First, we wanted to determine Campylobacter incidence, in particular C. concisus, in the pediatric population of Brussels suffering from AGE in a matched control population. The second objective was to genotype recovered C. concisus to highlight a possible difference between those from cases and controls.
MATERIALS AND METHODS
Patients 0 to 15 years of age, presenting with AGE to the pediatric emergency room of 2 university hospitals in Brussels, were recruited prospectively from May 2015 to October 2016. AGE was defined as a decrease in the consistency and/or an increase in the frequency of stools (≥3 in 24 hours), for less than 7 days. A stool sample was collected during the emergency room visit or within the 2 days after.
Controls, matched for age (±2 months), were prospectively recruited from outpatient general pediatric clinics from the same hospitals. Patients with current antibiotic treatment or AGE in the month after clinic visit, detected by a call 1 month later, were excluded.
Stools from cases and controls were analyzed for common bacteria (culture method for Salmonella, Shigella, Yersinia, Aeromonas, Arcobacter, enteropathogenic Escherichia coli), viruses (immunochromatographic methods for Norovirus, Rotavirus and Adenovirus) and parasites (microscopy).
To isolate Campylobacter, a Butzler medium (Thermofisher Scientific, Erembodegem, Belgium) was incubated at 42°C in microaerophilic atmosphere (CO2 10%, O2 5%, H2 0%) for 3 days. The filtration method6 was also performed, on Columbia agar containing 5% sheep blood (Becton Dickinson, Erembodegem, Belgium) using 0.45 µm pore size cellulose acetate filters (Porafil, Duren, Allemagne) from May 6, 2015, to May 26, 2016, (146 patients and 125 controls) and 0.60 μm pore size polycarbonate filters (Nuclepore, Whatman, Kent, United Kingdom) from May 26 to October 10, 2016 (38 patients and 51 controls), plates were incubated in microaerophilic atmosphere (CO2 10%, O2 5%, H2 0%) at 37°C for 10 days. All colonies suspected to be Campylobacter were identified by matrix-assisted laser desorption ionization-time of flight Mass Spectrometry using the Microflex LT and the IVD (in vitro diagnostic) 2.2 Biotyper database (Bruker, Bremen, Germany). C. concisus isolated during the study were genotyped using genomospecies (GS)-specific polymerase chain reaction (PCR).7
χ2 and Fisher exact test were used to compare pathogen prevalence in cases and controls. Univariate statistical analyses were done with Epi Info 7 version 126.96.36.199. Multivariate analyses using conditional logistic regression were performed with STATA 13 (StataCorp, College Station, TX).
Between May 2015 and October 2016, among 185 patients and 179 controls recruited in the 2 participating hospitals, 186 cases and 176 controls were analyzed for Campylobacter (Table, Supplemental Digital Content 1, http://links.lww.com/INF/D84). An enteropathogen was detected in 89 cases (48.1%) and 25 controls (14.0%). In cases, technical incidents led to missing results for bacterial culture (n = 1), viral studies (n = 3) and ova and parasite test (n = 4). In controls, technical issues led to missing results for Campylobacter cultures (n = 3), other bacteria (n = 3), viral studies (n = 1) and ova and parasite test (n = 2). C. jejuni was the most prevalent enteropathogen followed by Rotavirus, Adenovirus and Salmonella. Prevalence between cases and controls was not significantly different for other pathogens (Table, Supplemental Digital Content 2, http://links.lww.com/INF/D85), including for C. concisus, isolated from 6 cases and 4 controls. Codetection was present in 50% (3/6) of cases for C. concisus.
The multivariate analysis, adjusting for potential confounders, did not show any association between C. concisus and AGE (odds ratio = 0.89; 95% confidence interval: 0.16–4.45), unlike C. jejuni (odds ratio = 14.68; 95% confidence interval: 3.93–95.87). Nine of the 10 strains of C. concisus isolated were genotyped. One strain each from a control and a case belonged to genomospecies 2, 2 strains from controls and 3 strains from cases belonged to genomospecies 1 and 1 strain from a control and 1 from a case were positive for both genomospecies 1 and 2.
Our study does not provide evidence that C. concisus has an etiologic role in pediatric AGE and few C. concisus were found in both cases and controls. Campylobacter is confirmed as one of the most important enteric pathogens, with an incidence of 17.4% (32/184). C. jejuni was the major contributor to this result because no C. coli was found in cases. The incidence of C. concisus was really low and not significantly different between cases and controls, even when genotyping technique was performed.
Previous studies had concluded that C. concisus was more likely to be an opportunistic pathogen or part of the commensal microflora in the general population.8,9 The present work, even with particular attention paid to the selection of controls, supports this hypothesis. Even with a sizeable sample size, we were not able to find a significant difference. Given the 2.3% prevalence for C. concisus in the control group, a sample size of 7000 cases and controls would have been necessary, with 80% power. The diagnostic tests used for the isolation of C. concisus were a clear limitation. The filter-based culture method use, partly with a filter known to reduce the recovery of nontargeted bacteria, decreased our sensitivity.6 The gas mixture used for culture could not include hydrogen, which could contribute also to the low prevalence of C. concisus observed.
Molecular biology could have been useful in that context, but PCR methods detect bacterial DNA, whether the bacteria is alive or not, which can be confusing when studying stools for the presence of bacteria known to colonize the human oral cavity.5 Interestingly, even with molecular methods, others have failed to prove a pathogenic role for C. concisus.2,4
Another point against a pathogenic role for C. concisus is coinfection (Fig., Supplemental Digital Content 3, http://links.lww.com/INF/D86). C. concisus was isolated alone in 4 controls but, among the 6 cases with C. concisus, 3 were coinfected with confirmed pathogens: Salmonella, rotavirus and norovirus. Moreover, coinfection is likely underestimated because our study was done under routine laboratory conditions without molecular tools.
C. concisus is genetically diverse and genomospecies are thought to differ in their pathogenic roles.10 The genomospecies-specific PCR used in this study to type C. concisus had been used previously.7 Kalischuk et al10 found an increased proportion of GS1 in C. concisus in stools from healthy controls (4/5 strains) in comparison with an increase in the proportion of GS2 in diarrheic stools (12/17 strains) from Alberta (Canada). Other genotyping methods and virulence factors have been proposed, but each involved only a very small number of strains and none could demonstrate a clear difference between strains found in diarrheic versus healthy patients. The data obtained in the present work, despite its small sample size, suggests that both genomospecies are part of the commensal microflora or are, at most, opportunistic pathogens.
The major strength of our work is the similarity between controls and cases. The study could be considered as monocentric, because both participating hospitals are located in Brussels, serving the same low-income population. The generalizability of our results may thus be limited. Sample size, although sufficient for most AGE pathogens, did not allow us to draw firm conclusion on the pathogenic role of C. concisus considering its low general prevalence, similar in cases and controls. Finally, the specific diagnostic filtration technique used to isolate C. concisus could be optimized by changing the filters used6 and adding H2 to the growth atmosphere.
1. Nielsen HL, Ejlertsen T, Engberg J, et al. High incidence of Campylobacter concisus
in North Jutland, Denmark: a population-based study. Clin Microbiol Infect Off Publ Eur Soc Clin Microbiol Infect Dis. 2013;19:445–450.
2. Inglis GD, Boras VF, Houde A. Enteric campylobacteria and RNA viruses associated with healthy and diarrheic humans in the Chinook health region of southwestern Alberta, Canada. J Clin Microbiol. 2011;49:209–219.
3. Collado L, Gutiérrez M, González M, et al. Assessment of the prevalence and diversity of emergent campylobacteria in human stool samples using a combination of traditional and molecular methods. Diagn Microbiol Infect Dis. 2013;75:434–436.
4. Cornelius AJ, Chambers S, Aitken J, et al. Epsilonproteobacteria in humans, New Zealand. Emerg Infect Dis. 2012;18:510–512.
5. Kaakoush NO, Mitchell HM. Campylobacter concisus
- a new player in intestinal disease. Front Cell Infect Microbiol. 2012;2:4.
6. Nielsen HL, Engberg J, Ejlertsen T, et al. Comparison of polycarbonate and cellulose acetate membrane filters for isolation of Campylobacter concisus
from stool samples. Diagn Microbiol Infect Dis. 2013;76:549–550.
7. On SLW, Siemer BL, Brandt SM, et al. Characterisation of Campylobacter concisus
strains from South Africa using Amplified Fragment Length Polymorphism (AFLP) profiling and a Genomospecies-specific Polymerase Chain Reaction (PCR) assay: Identification of novel genomospecies and correlation with clinical data. Afr J Microbiol Res. 2013;7:1845–1851.
8. Van Etterijck R, Breynaert J, Revets H, et al. Isolation of Campylobacter concisus
from feces of children
with and without diarrhea
. J Clin Microbiol. 1996;34:2304–2306.
9. Engberg J, On SL, Harrington CS, et al. Prevalence of Campylobacter, Arcobacter, Helicobacter, and Sutterella spp. in human fecal samples as estimated by a reevaluation of isolation methods for Campylobacters. J Clin Microbiol. 2000;38:286–291.
10. Kalischuk LD, Inglis GD. Comparative genotypic and pathogenic examination of Campylobacter concisus
isolates from diarrheic and non-diarrheic humans. BMC Microbiol. 2011;11:53.