Abstract
Previous studies in patients without Type 2 diabetes suggest that brain hypo- and hypermetabolic regions may indicate risk for cognitive disorders. We aimed to study these brain glucose uptake patterns in Type 2 diabetes to assess cognitive disorder risk and improve personalized management. Six hyper- and three hypometabolic regions were obtained through statistical parametric mapping, with cerebellar vermis and right superior temporal gyrus being the most relevant areas, respectively. Such allowed identification of two phenotypes via k-means clustering: brain hypometabolic dominant (bU[-]) and hypermetabolic dominant (bU[+]). bU[-] displayed elevated markers of both Type 2 diabetes and cognitive disorders, specifically of secreted frizzled-related protein 1, a protein related to different neuronal pathologies. A classifier was developed (area under the curve = 0.84, true positive rate = 0.81 and true negative rate = 0.78) using a combination of biochemical features. Type 2 diabetes patients exhibit hypo- and hypermetabolic brain regions that phenotype into bU[-] and bU[+] by using the relationship between right superior temporal gyrus and cerebellar vermis, which defines the transition from one phenotype to the other. We suggest bU[-] patients are exposed to a higher risk of developing cognitive disorders based on the alteration of secreted frizzled-related protein 1 due to progressed type 2 diabetes, which can be identified using the proposed biomarker-based classification model.