In a prospective study of CD infection in hospitalized children conducted at 2 referral pediatric centers, nearly one-third of subjects whose stool was ELISA-positive for CD toxin A and B were negative for CD by culture. While asymptomatic excretion of CD is a known phenomenon in children, poor specificity of the ELISA toxin assay in pediatrics has not been previously appreciated.
To determine the underlying basis for the unexpectedly large numbers of CD ELISA toxin assay-positive / culture negative (ELISA+/cul-) stools in children.
Children were enrolled in the prospective study if their stool, submitted at the discretion of the attending pediatrician to the clinical microbiology laboratory, tested positive for CD toxin by ELISA (Meridian Premier Toxin A&B). ELISA toxin-positive samples were transferred to a research laboratory and cultured on pre-reduced CDBA plates with and without enrichment. If samples were negative for CD growth, they further were cultured after serial dilution to remove putative inhibitors. Prospectively collected clinical characteristics of children with ELISA+/cul- stools were compared with those with ELISA toxin-positive / culture-positive (ELISA+/cul+) stools using standard statistical tests. Pediatric stools that were ELISA+/cul- were further evaluated for presence of the organism by assaying for CD glutamate dehydrogenase (GDH) antigen (Wampole C. diff. Quik Check) and for the tcdB gene by PCR (BD GeneOhm). Stool also was cultured for C. sordellii, a previously reported cause of false-positive CD ELISA toxin assays.
99 children whose stool was CD ELISA toxin-positive stool were studied, of whom 35 (35%) were CD culture-negative. Children with ELISA+/cul- stool were significantly younger compared with those with ELISA+/cul+ samples (median (25,75 IQR) age .68 (.34,5.48) vs 3.56 (1.07,10.88) years, P < .003). ELISA+/cul- subjects were not different from those with ELISA+/cul+ samples by any other tested characteristic, namely, history of CD in the prior year (11% vs 8%), hospitalization in the prior 30 days (42% vs 38%), antibiotics in the prior 30 days (61% vs 72%), diarrhea on admission (31% vs 28%), presence of a co-morbidity (74% vs 80%), and positive presence of blood in the stool (33% vs 31%, all P > .30). 32 pediatric ELISA+/cul- stools (16 from the prospective study, 16 from clinical microbiology archives) were further evaluated for the presence of CD by GDH and PCR assay, of which only 2 (6%) were positive, both by both assays. 16/32 specimens were cultured for C. sordellii, and all were negative.
An unexpectedly high proportion of false-positive CD ELISA toxin assays was found among samples from pediatric patients, particularly in young subjects. These data raise concerns regarding the validity of the ELISA toxin assay alone to diagnose CD infection in this age group. The underlying basis for this phenomenon was not determined.