Epsilonproteobacteria in Humans, New Zealand

To the Editor: Cornelius et al. (1) addressed the potential of Campylobacter ureolyticus as an emerging pathogen by conducting a molecular study on 128 diarrheal specimens and 49 fecal samples from healthy volunteers. Reporting the identification of C. ureolyticus in 12 (24.5%) of 49 healthy volunteers, a number that they compared with our finding of 349 (23.8%) from Campylobacter spp.–positive samples (2), the authors concluded that C. ureolyticus species “are unlikely causes of diarrhea,” an assertion with which we take issue. 
 
This interpretation does not take into account that our screening involved 7,194 symptomatic patients: a sample size 40× greater than that of Cornelius et al. In this context, the likely carriage rate for C. ureolyticus is 1.15%. Also, our assay, which has a limit of detection in the picomolar range, is likely comparable with, if not greater than, that of Cornelius et al. (1). 
 
Accounting for variations in geographic location and detection methods, a detection rate of 24.5% in healthy volunteers (overall detection rate 14.7%) is high in contrast to our reported rate of 1.15%. One possible explanation for this discrepancy is that Cornelius et al. “did not specifically exclude volunteers who had had gastrointestinal disturbances in the 10 days before sampling,” Campylobacter can be shed in feces for <4 weeks after infection. Also, Cornelius et al. (1) noted the possibility of “genetically distinct but phenotypically indistinguishable genomospecies differing in their pathogenic potential” to account for the presence of the emerging pathogen C. concisus in healthy volunteers and patients with diarrheal illness. This may also apply for C. ureolyticus. 
 
We reported a strong seasonal prevalence of C. ureolytcius and a bimodal age distribution (2). The lack of any related details from Cornelius et al. may undermine their reported detection rates. These factors strongly suggest that the statement, “these species are unlikely causes of diarrhea,” should, at the very least, be taken under advisement.


Epsilonproteobacteria in Humans, New Zealand
To the Editor: Cornelius et al. (1) addressed the potential of Campylobacter ureolyticus as an emerging pathogen by conducting a molecular study on 128 diarrheal specimens and 49 fecal samples from healthy volunteers. Reporting the identifi cation of C. ureolyticus in 12 (24.5%) of 49 healthy volunteers, a number that they compared with our fi nding of 349 (23.8%) from Campylobacter spp.-positive samples (2), the authors concluded that C. ureolyticus species "are unlikely causes of diarrhea," an assertion with which we take issue.
This interpretation does not take into account that our screening involved 7,194 symptomatic patients: a sample size 40× greater than that of Cornelius et al. In this context, the likely carriage rate for C. ureolyticus is 1.15%. Also, our assay, which has a limit of detection in the picomolar range, is likely comparable with, if not greater than, that of Cornelius et al. (1).
Accounting for variations in geographic location and detection methods, a detection rate of 24.5% in healthy volunteers (overall detection rate 14.7%) is high in contrast to our reported rate of 1.15%. One possible explanation for this discrepancy is that Cornelius et al. "did not specifi cally exclude volunteers who had had gastrointestinal disturbances in the 10 days before sampling," Campylobacter can be shed in feces for <4 weeks after infection. Also, Cornelius et al. (1) noted the possibility of "genetically distinct but phenotypically indistinguishable genomospecies differing in their pathogenic potential" to account for the presence of the emerging pathogen C. concisus in healthy volunteers and patients with diarrheal illness. This may also apply for C. ureolyticus.
We reported a strong seasonal prevalence of C. ureolytcius and a bimodal age distribution (2). The lack of any related details from Cornelius et al. may undermine their reported detection rates. These factors strongly suggest that the statement, "these species are unlikely causes of diarrhea," should, at the very least, be taken under advisement.   (1), a major aspect of our study (2) was to compare epsilonproteobacterial populations in healthy persons and those who have diarrhea. We have not examined as many diarrheal samples as Bullman et al. (3). However, in contrast with their study, we have examined samples from persons with no evident disease manifestations. Because the presence of an agent during disease is not proof of causation, we believed that a baseline for comparison was needed. Campylobacter ureolyticus was found in a greater proportion of samples from healthy persons (24%) than samples from persons who had diarrhea (11%) (p = 0.041, by χ 2 test).
Samples from healthy persons were tested on 2 occasions: 18 samples in September 2007 (New Zealand spring) at the Institute of Environmental Science and Research, Christchurch, in the workplace, and 31 samples in June 2009 (New Zealand winter), at Christchurch Hospital under the guidance of a clinician. We have no reason to believe any of the workplace samples were provided when volunteers had diarrhea, particularly considering our workplace guidelines and staff characteristics. In each testing round, 6 fecal samples had positive test results for C. ureolyticus. These periods equate to the peak and trough periods described by Bullman et al. (3). We were unable to provide many details regarding sampling in our paper because of space constraints.
Considering our baseline comparisons of healthy persons with those who had diarrhea, we affi rm our con-clusions are reasonable and that C. ureolyticus is an unlikely cause of acute diarrheal disease. Similarly, C. ureolyticus (previously classifi ed as Bacteroides ureolyticus) has been detected in patients with Crohn's disease and in controls (4). However, different subtypes or undescribed subspecies may be pathogenic: some strains exhibit certain pathogenic characteristics in vitro (5) and others yield amplifi ed fragment length polymorphism profi les that are visually quite distinct from others (6). Host factors also cannot be discounted.
Stephen L.W. On and Angela J. Cornelius