These factors complicate study of the gastrointestinal microbiota, and direct comparison of results between studies may be problematic. However, comparison of the composition of the microbiota between groups of animals within a study such as ours is not subject to as many of these limitations, and is likely to generate more meaningful results. Our results showed that the Torin 1 presence of insulin-treated diabetes mellitus in cats does not affect faecal microbiota composition, as evaluated by the UniFrac distance metric or by comparison of relative abundances of predominant bacterial taxa identified by sequencing of the 16S rRNA gene. We were therefore unable to replicate the results of Serino et al. who described a decreased proportion of Firmicutes in mice with type 2 diabetes mellitus, or Larsen et al. who reported a similar finding in type 2 diabetic men, in cats with diabetes mellitus. It is possible that the inability to identify a difference in microbiota composition between diabetic and non-diabetic cats could have been due to the relatively small sample size in this study; however, previous studies that have reported compositional differences of the microbiota associated with obesity, type 2 diabetes and type 1 diabetes have studied a similar number of or fewer individuals, making type II error unlikely. An additional consideration is that all diabetic cats in this study were treated with insulin, this being standard therapy for feline diabetes mellitus. Whether or not exogenous insulin can alter microbiota composition and/or obscure diabetesassociated changes in microbiota composition is unknown, however future studies could explore this issue by studying diabetic cats at the time of diagnosis, prior to commencement of insulin therapy. Compositional analysis of the microbiota, as undertaken in this study, may overlook the complexities of microbial communities in vivo. In a recent study, faecal microbiota of children was examined at several time points up to three years of age, and the microbiota composition of children who developed anti-islet cell antibodies was compared with children who remained antibody-free. No differences in microbiota composition, relative proportions of bacteria at genus level, or diversity were noted between groups. However when a microbial correlation network was constructed, a significant difference was noted in microbial interaction networks between the two groups of children. It was concluded that despite an absence of compositional differences, microbial interaction networks were compromised in children who developed anti-islet cell antibodies. This study demonstrates that disease-associated alterations of the faecal microbiota may not necessarily be discernible as quantitative compositional changes; and that consideration of intramicrobiota relationships may afford a more comprehensive assessment of the microbiota. Importantly, failure to identify compositional differences of faecal microbiota between diabetic and non-diabetic cats does not exclude the possibility of functional differences of the microbiota in affected individuals. Host metabolic effects may not be entirely predictable by a particular microbiota composition.