Risk factors for cognitive decline in older people with type 2 diabetes

Abstract

People with type 2 diabetes are at increased risk of age-related cognitive impairment. Previous literature has focused on case-control studies comparing rates of cognitive impairment in patients with and without diabetes. Investigations of potential risk factors for cognitive impairment (including those with increased prevalence in diabetes, such as macrovascular disease, and diabetes-specific factors such as hypoglycaemia) in study populations consisting exclusively of patients with type 2 diabetes have been largely neglected. Moreover, previous studies have failed to take advantage of the extensive characterisation and prospective nature of longitudinal cohort studies to investigate the relative predictive ability of a wider range of potential risk factors for cognitive decline. Using data from the prospective Edinburgh Type 2 Diabetes Study (ET2DS) the present thesis aimed (i) to determine associations of cognitive decline with macrovascular disease and with severe hypoglycaemia, and (ii) to compare a wider range of potential risk factors in their ability to predict cognitive decline. In 2006/2007, 1066 patients with type 2 diabetes (aged 60 to 75 years) attended the baseline ET2DS clinic and 831 returned for the follow-up at year 4. Subjects were extensively characterised for risk factor profiles at baseline, and at year 4 for incidence of severe hypoglycaemia. Socioeconomic status was estimated using postcode data. Scores on seven tests of age-sensitive ‘fluid’ cognitive function, which were administered at baseline and at year 4, were used to derive a general cognitive component (‘g’). A vocabulary-based test, administered at baseline, estimated pre-morbid ability. Findings are reported in three parts. 1.) Macrovascular disease and cognition: Subjects with higher levels of biomarkers indicative of subclinical macrovascular disease, including plasma N-terminal pro-brain natriuretic peptide and carotid intima-media thickness, had significantly steeper four-year cognitive decline, independent of traditional cardiovascular risk factors, stroke, socioeconomic status and estimated pre-morbid cognitive ability. For ankle-brachial pressure index, the association fell just short of statistical significance. Effect sizes were overall modest, with fully adjusted standardised beta coefficients ranging from 0.06 to -0.12. Little evidence was found for associations of the symptomatic markers of macrovascular disease with four-year change in cognitive function that was independent of participants’ pre-morbid ability and socioeconomic status. 2.) Severe hypoglycaemia and cognition: Subjects with lower cognitive ability at baseline were at two-fold increased risk of experiencing their first-ever incident severe hypoglycaemia during follow-up. The rate of four-year cognitive decline was significantly steeper in those exposed to hypoglycaemia compared with hypoglycaemia-free participants, independently of cardiovascular risk factors, microand macrovascular disease and of estimated pre-morbid cognitive ability. Effect sizes again were overall modest (Cohen’s d = 0.2 to 0.3 for statistically significant differences in four-year cognitive decline between subjects with and those without hypoglycaemia, following multivariable adjustment) 3.) Consideration of a wider range of risk factors and cognition: A stepwise linear regression model including a total of 15 metabolic and vascular risk factors identified inflammation, smoking and poorer glycaemic control (in addition to some of the subclinical markers of macrovascular disease) as predictive of a steeper four-year cognitive decline. Other traditional cardiovascular risk factors, diabetic retinopathy, clinical macrovascular disease and a baseline history of severe hypoglycaemia were not included in this model. The interpretation of the latter finding is limited, however, by the fact that the stepwise regression procedure may exclude true predictors from a model when they correlate with already included risk factors. This thesis has demonstrated associations of later-life cognitive decline in people with type 2 diabetes with markers of subclinical macrovascular disease and poor glycaemic control (including hypoglycaemia) as well as other cardiometabolic risk factors (inflammation, smoking). Findings suggest that associations are relatively weak and complex due to inter-relationships amongst risk factors, and indicate a role of pre-morbid ability and socioeconomic status (which as risk factors are difficult to modify) in the relationships of risk factors with cognitive decline. Future research including case-control studies to compare risk factor associations between people with type 2 diabetes and non-diabetic older adults and randomised controlled trials to evaluate potential causal effects of individual modifiable risk factors on cognitive decline, will help to evaluate the mechanisms underlying the observation that people with type 2 diabetes are at risk of cognitive impairment in later life.