The amount of research done on the human microbiome (and those of other creatures) has increased hugely over the last twenty years or so, and it seems that hardly a week goes by without some news item about how our microbial hangers-on affect some part of human physiology in health and disease. Overall, I think that it’s a good thing that we’re paying attention to this aspect of biology, because the the (very large) number of microorganisms that live in and on our bodies really do seem to have a number of effects that we hadn’t been aware of. Gut health, metabolism of foodstuffs and pharmaceuticals, and even entire signaling networks within the body are all tied in to these things, and we’re clearly not going to have anything like a complete idea of what’s going on in the human body without considering the parts that aren’t exactly human.
That said, it’s a hard field to work in. Microbial communities are quite variable and not always easy to characterize, and the different species affect each other as well as affecting the rest of the body. This new paper and this one point out yet another difficulty: getting accurate samples in the first place. The authors in both papers have tried all sorts of sample-handling variations, and it looks like they have had trouble finding any that don’t change the composition of the microbial samples themselves. Both papers investigated the two commercially available stool sample kits (OMNIgene and Zymo), and found that the latter was much more sensitive to temperature variations on storage. And both kits changed the absolute levels of various bacteria types: the OMNIgene-preserved samples had significantly higher amounts of Bacteroidetes species as compared to preservative-free controls, while the Zymo-preserved ones had significantly lower amounts. The second paper also finds that the method used for cell disruption can significantly affect the ribosomal RNA reads used to characterize the bacterial species as well.
The authors of the first paper suggest that researchers in the field should also be measuring total bacterial load in their samples and monitoring that for signs of variability in their sample handling and the second paper also suggests that people should standardize on a particular number of PCR cycles (25), because that can also change things. These effects, say the authors, can help explain the widely varying literature results in human microbiome studies. Let’s hope so! This field has suffered from a lot of failure-to-reproduce, and if ironing out these differences can help, then the sooner the better. But a pessimist might note that there have been such calls before! I’m sure that standards are higher than they were earlier in the field’s history, but there seems little doubt that they still have more room to improve. Can human microbiome papers really be compared under current conditions, and can overarching conclusions really be drawn by doing so? If not, how many more times will researchers have to be reminded about these problems?