Timing in Collection of Stool Samples

See allHide authors and affiliations

Science  18 Nov 2005:
Vol. 310, Issue 5751, pp. 1118
DOI: 10.1126/science.310.5751.1118

We read with great interest the Report “Diversity of the human intestinal microbial flora” by P. B. Eckburg et al. (10 June, p. 1635; published online 14 Apr.). We applaud the authors for advancing this important field by undertaking comprehensive 16S rDNA sequencing to describe the composition of the microbiota in stools and at six sites of the colon in each of three human volunteers. The analyzed 13,355 prokaryotic ribosomal RNA gene sequences represent the largest 16S rDNA microflora data set reported to date in any species.

On the basis of their analysis, the authors suggest that “differences between stool and mucosa community composition” exist. We question the validity of this conclusion, based on our own observation that stool microflora composition can vary significantly in stool samples collected before and after a colonoscopy (1). The authors compare microflora composition in colon biopsy samples obtained during colonoscopy with a stool sample collected a month afterwards. The authors acknowledge potential problems with their interpretation because of the delayed stool collection, but a rationale for collecting delayed stool samples is not given.

This Report has significantly expanded our knowledge of the diversity of the intestinal microflora in a few subjects. However, if we ever want to correlate microflora composition with health or disease, we will have to design studies aimed at understanding the variation in the microflora composition in a large cohort of human subjects.


  1. 1.


Our large-scale comparative analysis of bacterial and archaeal 16S rDNA sequences in the colon and stool revealed significant intersubject variability and patchy heterogeneity among the colonic mucosal bacterial populations. Regarding the statistical differences we reported between stool and adherent mucosal populations, subjects B and C harbored different bacterial populations in their colonic mucosa compared with their stool samples collected 4 weeks later, while the mucosal populations in subject A were subsets of the population observed in stool collected 4 weeks after colonoscopy. Each subject's stool community was more similar to the communities of their own mucosal samples than to any community from a different subject.

We acknowledged that the statistically significant difference between the bacterial composition of the stool and colonic mucosa may have been due to the 4-week delay in stool collection after colonoscopy. The collection of stool was not originally planned in the large Canadian population-based case control study from which the control subjects were selected. For this study of three healthy subjects, from whom the colonic tissue biopsies had already been collected, we chose to obtain stool samples 1 month after colonoscopy when each subject had full recovery of stable, baseline bowel function. Despite the original study design, we agree with Mai et al. that stool samples before endoscopy may provide more reliable representations of the steady-state stool bacterial population. However, a controlled comparative study is needed to reveal the degree to which stools are microbiologically dissimilar at various time intervals before and after bowel cleansing. A small study using denaturing gradient gel electrophoresis has suggested that the bacterial composition in colonic mucosal biopsies differs significantly from that in stool collected prior to the procedure (1), supporting our conclusions that significant differences exist between these microbial communities. Future studies should address these issues with high-resolution molecular methods and a greater number of subjects.


  1. 1.

Navigate This Article