Background: Plasma amylin is positively associated with cognitive function in humans. Amylin treatment improves memory in Alzheimer's mouse models. However, the relationship between plasma amylin, diabetes and cognition is not clear. Objectives: In this study we examined the concentration of plasma amylin, its relationship with diabetes and cognition. Material and Method: A cross-sectional, homebound elderly population with data of plasma amylin under fasting condition and cognitive measurements was used. Results: We found that subjects with a long and chronic duration of diabetes were more likely to take insulin treatment and have reduced secretion of amylin. Compared to non-diabetics, diabetic subjects without insulin treatment had a higher concentration, but those with insulin treatment had a lower concentration, of plasma amylin [median (Q1, Q3): 20 (11.0, 36.2) vs. 25.2 (13.2, 50.6) vs. 15.0 (4.9, 33.8), p< 0.0001]. In the whole sample vs. in the absence of diabetes, plasma amylin was positively associated with logical memory delayed recall (beta= + 0.61, SE= 0.25, p= 0.02 vs. beta=+ 0.80, SE= 0.33, p= 0.02) and block design (beta=+ 0.62, SE= 0.24, p= 0.009 vs. beta=+ 0.93, SE= 0.31, p= 0.003), and negatively associated with Trailmaking A scores (beta=-6.21, SE= 1.55, p< 0.0001 vs. beta=7.51, SE= 1.95, p= 0.0001) and Trailmaking B (beta=-4.32, SE= 2.13, p= 0.04 vs. beta=-5.86, SE= 2.73, p= 0.04). All these relationships disappeared in the presence of diabetes regardless the treatment. Conclusion: This study suggests that secretion of amylin by pancreas compensates and then deteriorates depending on the duration of diabetes. Amylin's activities for cognition are impaired in the presence of diabetes.
