Detection of Campylobacter in stool and determination of significance by culture, enzyme immunoassay, and PCR in developing countries

Abstract

Campylobacter is a common bacterial enteropathogen that can be detected in stool by culture, enzyme immunoassay (EIA), or PCR. We compared culture for C. jejuni/C. coli, EIA (ProSpecT), and duplex PCR to distinguish Campylobacter jejuni/C. coli and non-jejuni/coli Campylobacter on 432 diarrheal and matched control stool samples from infants in a multisite longitudinal study of enteric infections in Tanzania, Bangladesh, and Peru. The sensitivity and specificity of culture were 8.5% and 97.6%, respectively, compared with the results of EIA and 8.7% and 98.0%, respectively, compared with the results of PCR for C. jejuni/C. coli. Most (71.6%) EIA-positive samples were positive by PCR for C. jejuni/C. coli, but 27.6% were positive for non-jejuni/coli Campylobacter species. Sequencing of 16S rRNA from 53 of these non-jejuni/coli Campylobacter samples showed that it most closely matched the 16S rRNA of C. hyointestinalis subsp. lawsonii (56%), C. troglodytis (33%), C. upsaliensis (7.7%), and C. jejuni/C. coli (2.6%). Campylobacter-negative stool spiked with each of the above-mentioned Campylobacter species revealed reactivity with EIA. PCR detection of Campylobacter species was strongly associated with diarrhea in Peru (odds ratio [OR] = 3.66, P < 0.001) but not in Tanzania (OR = 1.56, P = 0.24) or Bangladesh (OR = 1.13, P = 0.75). According to PCR, Campylobacter jejuni/C. coli infections represented less than half of all infections with Campylobacter species. In sum, in infants in developing country settings, the ProSpecT EIA and PCR for Campylobacter reveal extremely high rates of positivity. We propose the use of PCR because it retains high sensitivity, can ascertain burden, and can distinguish between Campylobacter infections at the species level.

FIG 1

Quantity of C. jejuni/C. coli (cadF) and Campylobacter species (16S rRNA) by real-time PCR for all samples categorized by EIA result (n = 432 samples). Samples with Campylobacter jejuni/C. coli present would be expected to fall along the diagonal band running from the bottom left to top right, suggesting a similar level of detection of the two Campylobacter PCR targets.

FIG 2

Quantity of Campylobacter species (16S rRNA) by real-time PCR for case and control samples for each site. Among Campylobacter-positive samples, real-time PCR C_qs are shown for each site as the median, interquartile range, and range. The burden of asymptomatic Campylobacter infection was statistically significantly lower in Peru than in the other two sites by the Mann-Whitney U test. No other two-way comparisons between sites were statistically significant. The difference between all sites was statistically significant (P = 0.002, Kruskal-Wallis one-way ANOVA).