Using the Gut Microbiome Genome to Predict Type 2 Diabetes

Ali A. Torkamani, PhD


July 31, 2013


A major issue with the development and evaluation of predictive models for T2D is the need to differentiate between T2D incidence vs prevalence.

For example, traditional diabetes scores, such as the FINDRISC score, utilize predictors that are typically comorbid with T2D, such as elevated body mass index (BMI). Yet, it is not clear what the real clinical utility is in predicting whether an individual with elevated BMI is destined to develop T2D. It seems unlikely that the recommendations of a physician for an individual with elevated BMI would differ substantially if that individual were or were not destined to develop T2D. Moreover, such factors as BMI, feeding behavior, and family history are themselves at least in part genetic, absorbing the predictive power of genetic markers in cases where T2D onset is already well on its way.

In other words, these models and evaluation scenarios capture the contribution of genetic predictors to disease incidence at a point in time during the development of T2D where a prospective prediction seems to be of limited value. The onset of the disease, while perhaps not yet fitting the diagnostic criteria for overt T2D, has already begun.

A similar criticism can be raised for the utility of microbiome-based predictors of T2D, especially if the abundance of bacterial species in T2D individuals is an effect rather than a cause of the disease. Although the findings in this study are interesting, it's not clear that placing an individual at a more precise phase in the progression from metabolic syndrome to T2D on the basis of the microbiome profile is of any significant clinical utility.

Ideal genetic prediction models have the potential for predicting the incidence of T2D before the appearance of any preclinical symptoms. Although one may argue that physician recommendations would not differ for an individual with high vs low genetic risk of T2D -- after all, everyone should eat well and exercise -- perhaps the real power of a genetic predictor lies in its utility as a motivational tool to influence good behaviors. Moreover, if knowledge of high genetic risk and gene/environment interactions is available, then behavioral change may be both more powerfully motivated and more specifically directed to avoid the most high-risk behaviors for the individual in question. Finally, there is limited but captivating evidence to suggest that influencing the microbiome profile of an individual can alter glucose tolerance.[5] Thus, there does appear to be potential for the use of microbiome genetic profiling in preventing, rather than predicting the prevalence of, T2D.