Lately, there’s been another very interesting proposal. This has come from Leif Groop and his associates. So, what Leif Groop has done here is to take the Swedish diabetic population, which is mainly comprised, as we know, of individuals with type 2 diabetes, and then tried to analyse this population on the basis of some simple factors like, for instance, the presence of antibodies, a proxy for insulin function and insulin sensitivity (the HOMA2-IR and HOMA2-B), the age of onset and the body mass index at the time of the onset. And running through the population, using these elements here, it was possible to identify five different clusters, which have been illustrated here. So, these clusters are mainly characterised by the presence of autoimmunity, which is much closer to type 1 diabetes. One cluster is mainly associated with severe impairment in insulin secretion, another cluster by severe impairment in insulin sensitivity, while another one is associated with mild obesity and another is mainly associated with age at the time of onset. Now, what is very interesting about this approach is, first of all, that it has been replicated in different populations. And what you can see is that it doesn’t matter which population you analyse; the distribution of these five clusters is very similar across all of them, thereby suggesting that this cluster analysis is sufficiently reliable.