Comparing the Metagenomic Performance of Stools Collected from Custom Cards and 95% Ethanol in Epidemiologic Studies

Background: Stool cards have been used for microbiome assessment in epidemiologic studies. Methods: We compared shotgun metagenomic sequencing from 32 participants who self-collected stool samples from the same bowel movement using a custom stool card versus a collection tube with 95% ethanol fixative in the Nurses’ Health Study II. We evaluated the agreement between methods at both the whole-community and individual species levels. To contextualize the comparison for disease association studies, we assessed the performance of the two collection methods for differentiating colorectal cancer–associated taxa. Results: Overall, metagenomes from cards and 95% ethanol were highly correlated within individuals. No difference was found in α diversity and only ∼1% of variation in β diversity was explained by the collection method. At the species level, although the relative abundances were highly correlated between card and ethanol sample pairs (Spearman rho ¼ 0.96), 10 (of 239) species showed a differential abundance in paired samples, including overrepresentation of Escherichia coli and underrepresentation of three Streptococcus species in cards compared with ethanol. Among a set of 99 colorectal cancer–associated species, four showed differential abundances between collection methods; however, this number was consistent with that expected by chance. Conclusions: Metagenomic sequencing using stool samples self-collected using stool cards or 95% ethanol yielded largely consistent results although differential abundances were observed for a small number of individual species.