Get Newsletter
AlzRisk Risk Factor Discussion
Back

Risk Factor:
Risk Factor Type: Chronic disease, Metabolic
Current Understanding:
The tables below summarize results from observational studies of the relationship between diabetes mellitus and measures of pre-clinical diabetes with AD and total dementia. Overall, the evidence is consistent with an association between diabetes diagnosis and increased risk of AD, suggesting that measures to prevent diabetes -- such as exercise, weight reduction and diet control -- will likely provide some protective benefit. Preliminary studies examining the risk of AD in association with measures of pre-clinical diabetes seem to specifically link impaired glucose tolerance with increased AD risk, but further research is needed to better characterize the relationship between these measures and AD. While standard glucose control is critical to prevent microvascular and macrovascular complications of diabetes, and may also be beneficial for cognitive outcomes, the effects of tighter glucose control regimens on AD require further study. For a review of the putative mechanisms by which diabetes may influence AD risk and detailed commentary on interpreting the findings below in a broader context, please view the Discussion.
Literature Extraction: Search strategy  * New *
Last Search Completed: 26 June 2012 - Last content update released on 1 Nov 2012

Risk Factor Overview

Cite as:

Sajeev G*, Weuve J*, McQueen MB, Blacker D. "Diabetes." The AlzRisk Database. Alzheimer Research Forum. Available at: http://www.alzrisk.org. Accessed [date of access]*.

*contributed equally

* * *

Introduction

The tables in the Risk Factor Overview summarize the results from a number of studies of the association between diabetes mellitus and Alzheimer disease (AD). Despite some inconsistent results, the evidence overall is consistent with an association between diabetes diagnosis and increased risk of AD. More recent work has investigated incident AD in relation to measures of pre-clinical diabetes, and the results are less clear. The exposures assessed have differed across studies, with some examining "prediabetes", “hyperinsulinemia” and "hyperglycemia" as categorical constructs, and others investigating levels of glucose, insulin and insulin resistance either as continuous or categorical variables. An overall view of this limited evidence suggests that prediabetic states may also be associated with increased AD risk. The results of these studies also seem to more strongly link impaired glucose tolerance, rather than elevated fasting glucose, to increased AD risk. The few investigations of insulin suggest a U-shaped association between fasting insulin levels and AD risk.

Mechanisms of Action

Diabetes mellitus is a disorder characterized by elevated blood glucose resulting from concomitant deficiencies in insulin production, in response to insulin, or both. The two most common forms of diabetes are type 1—an autoimmune condition in which the β cells of the pancreas are destroyed and thus no longer produce insulin—and type 2—in which the peripheral response to insulin is impaired, leading to increases in insulin production (and, eventually in some cases, reduced capacity to produce insulin). Many complications are associated with diabetes including macrovascular (e.g., ischemic heart disease and stroke) and microvascular disease (e.g., kidney disease, white matter disease of the brain). Hyperglycemia is also one part of the "metabolic syndrome," a cluster of co-occurring factors which increase the risk of cardiovascular disease and stroke. Given the established link between vascular disease and the pathogenesis and/or progression of dementia and AD [1, 2], the co-occurrence of diabetes with these factors likely contributes to a relationship between diabetes and dementia risk.

The direct effects of blood glucose on the brain may also partially explain the observed link between diabetes and AD/dementia. Chronic hyperglycemia results in increased oxidative stress and increased generation of advanced glycation end-products, two factors that appear to contribute to generalized atrophy and microvascular changes in the brain [3]. For type 1 diabetes, an additional potential mechanism is the neurologic damage caused by the frequent -- and, sometimes severe -- hypoglycemic episodes associated with tight glucose control. It is less clear whether the typically infrequent mild hypoglycemic episodes of type 2 diabetes might affect cognitive outcomes in this way [4].

Normal levels of insulin are integral to the physiology of learning and memory. While acute elevations of brain insulin may have beneficial effects on cognition, the chronic peripheral hyperinsulinemia resulting from prolonged insulin resistance appears to result in reduced brain insulin, which has been linked to accumulation of amyloid beta (Aβ) protein. In the brain and in the periphery, decreased insulin impairs the clearance of amyloid beta (Aβ) protein, which ultimately elevates Aβ levels in the brain. Decreased brain insulin reduces the release of intracellular Aβ to extracellular compartments where clearance occurs, and also reduces brain levels of insulin degrading enzyme, which is involved in clearance of Aβ. Increased peripheral insulin also decreases peripheral clearance of Aβ, further increasing Aβ levels in the brain. Decreased brain insulin also increases the hyperphosphorylation of tau protein leading to its aggregation in tau tangles, one of the characteristic lesions of AD. See Cholerton et al. for more details on the mechanisms linking insulin resistance and AD [5].

Methodological Issues

Exposures

The tables report on diabetes mellitus diagnosis, hyperglycemia, hyperinsulinemia and pre-diabetes as categorical constructs, and glucose, insulin and insulin resistance as either continuous or categorical variables. Each of these measures has inherent advantages and disadvantages in terms of capturing the aspects of diabetes and its subclinical precursors that may be relevant to AD risk.

Diabetes. Most papers evaluated diabetes diagnosis, although research on preclinical disease states and biomarkers of disease is becoming increasingly common. Very few reports listed here distinguished type 1 from type 2 diabetes. However, as type 2 diabetes accounts for the overwhelming majority of diabetes among older adults (due to its overall higher prevalence, the increased prevalence with age, and the shorter life expectancy among type 1 diabetics), the findings of these studies, by and large, likely pertain to type 2 diabetes.

Overall, the studies differed somewhat in their assessment of participants' diabetic status. Only a few administered their own blood glucose tests to diagnose participants. In most studies, investigators classified participants based on self-reports or medical records indicating diabetes diagnosis and/or use of insulin or oral hypoglycemic agents. Although less sensitive, these methods likely have good specificity, and their use should only attenuate effect estimates, provided underdiagnosis of diabetes is unrelated to the subsequent diagnosis of AD.

If the effects of diabetes-related factors on AD risk are cumulative, it is important to account for duration, severity and control of diabetes, which most studies have not been able to do. Although glycosylated hemoglobin (HbA1c) (a measure of glucose control) is becoming more commonly used in epidemiologic studies, it has not yet been studied in relation to AD incidence.

It may also be important to examine whether the relationship with AD differs by age of diagnosis, in case having diabetes in a specific time window (e.g., mid-life) is etiologically relevant for AD risk. There appears to be evidence of such an age-dependent association for blood pressure [6] and obesity [7], two risk factors related to diabetes. Specifically, mid-life hypertension and obesity have been linked to increased AD risk, while late-life hypertension and obesity are associated with decreased AD risk. However, most studies of diabetes have not had sufficiently detailed diagnosis information to determine whether the association between diabetes and AD is similarly age-dependent.

Finally, a classification system based solely on diabetes diagnosis also ignores the possibly elevated AD risk among “pre-diabetics” (see below). Such individuals would be classified as non-diabetics, potentially biasing any adverse association towards the null if pre-diabetes also increases AD risk.

Pre-diabetes. Pre-diabetes refers to an intermediate state of hyperglycemia in which blood glucose levels are elevated but are still lower than diagnostic thresholds for diabetes [8]. The few studies in our database that explored this exposure defined pre-diabetes following World Health Organization (WHO) criteria [9]. The American Diabetes Association has similar criteria [10]. Both sets of criteria distinguish between impaired fasting glucose (elevated fasting glucose) and impaired glucose tolerance (elevated 2-hour postload glucose).

The term pre-diabetes actually encompasses three states: isolated impaired fasting glucose (i-IFG), isolated impaired glucose tolerance (i-IGT), and combined IFG and IGT (IFG/IFT). While both i-IFG and i-IGT signal increased risk of progression to diabetes, these two states are metabolically distinct, have different suspected etiologies, and are hypothesized to progress to different type 2 diabetes phenotypes [11]. Distinguishing between these pre-diabetic states may be important, as differences in their etiology and pathophysiology may be relevant to AD incidence. For instance, although insulin resistance is characteristic of pre-diabetes, the sites of insulin resistance differ between i-IFG, i-IGT and IFG/IGT. IFG/IGT is characterized by both hepatic and muscle insulin resistance, i-IFG by specifically elevated hepatic but mostly unaltered muscle insulin resistance, and i-IGT by markedly elevated muscle but only slightly elevated hepatic insulin resistance [12]. Given these differences and the hypothesized mechanistic importance of increased peripheral insulin resistance, the associations of i-IGT and IFG/IGT with AD would be expected to be larger than that for i-IFG. Between-study differences in the prevalences of these underlying states could also partially explain inconsistent results between reports that analyze pre-diabetes or diabetes as binary diagnostic categories.

Glucose. Distinguishing between pre-diabetic states also helps to emphasize that different blood glucose tests (e.g., fasting, postload) may identify metabolic phenotypes with different levels of AD risk. Blood glucose diagnostic thresholds for diabetes were developed based on the markedly increased prevalence of retinopathy — a specific complication of diabetes — above those thresholds [10]. Above the diagnostic thresholds for diabetes, fasting and post-load glucose both signal similarly increased risk of retinopathy, but these measures are not identical with respect to predictions of other consequences of diabetes. For instance, post-load glucose is a better predictor of cardiovascular disease than fasting glucose [13], and similar differences may hold for prediction of AD [14]. Therefore, separate evaluations of these different measures across the range of their values will likely be necessary to more completely characterize the relationship between glucose and AD.

Other exposures. In general, hyperglycemia was defined according to fasting glucose level and use of anti-diabetic medication, as specified in the National Cholesterol Education Program Adult Treatment Panel III (NCEP ATP III) criteria [15] for metabolic syndrome. The designation of hyperglycemia was applied to qualifying participants irrespective of their diabetes diagnosis. A few studies examined participants’ fasting insulin levels and insulin resistance in relation to AD risk. Insulin resistance was estimated from fasting insulin and glucose levels, using the homeostatic model assessment (HOMA) technique [16], which has been well-validated against the gold standard hyperinsulinemic-euglycemic clamp method [17]. There have been no studies of the association between HbA1c and AD incidence.

Design and Analysis

Adjustment for confounding. Diabetes frequently occurs as part of "the metabolic syndrome," which also includes hypertension, hypercholesterolemia, elevated triglycerides, and obesity [18]. Each of these additional conditions markedly increases the risk of cardiovascular disease among diabetics [19]. Since vascular disease is associated with increased AD risk, it is likely that vascular factors partially explain the association between diabetes-related exposures and AD. Some studies elect to adjust for these potential vascular intermediates, which may lead to an underestimation of the overall effect of diabetes on AD risk. Some papers reported separate results with and without such adjustments, generally with little impact on the findings: the AlzRisk tables typically show results that were maximally adjusted, except for potential vascular intermediates.

Selection bias. The manner in which individuals are selected into a study cohort and subsequently into an analysis sample can produce biased results. Diabetics younger than 60 years old have at least double the mortality rate of nondiabetics of the same age [20], meaning that fewer diabetics survive to the age of eligibility for enrollment into studies of AD. Therefore, the diabetic participants in these studies represent a selected subset of individuals, likely with better-controlled, less severe or more recently acquired diabetes and better overall health. Further, death, worsening health and the burden of study participation contribute significantly to the selective depletion of diabetic participants from study cohorts over time. The mortality rate among diabetics aged 60-79 is 30% to 148% higher than among non-diabetics of the same age [20]. This pattern of selective attrition can be seen in longitudinal studies of cognitive decline and AD. For instance, a history of diabetes was predictive of greater attrition due to death among participants over age 65 in the Chicago Health and Aging Project [21]. Strong predictors of AD such as increasing age and worsening cognition have been also associated with a greater likelihood of attrition for reasons other than death in longitudinal studies of older adults [22, 23]. Consequently, the subset of participants remaining for analysis is likely to be healthier and have lower risk of AD than those initially selected. More important, remaining participants with diabetes may have lower-than-expected risk for AD, meaning that the increased risk of AD associated with diabetes is likely to be underestimated even further.

Effect Modification. In general, modifiers of the diabetes-AD link have not been studied in much detail. A few studies examined whether the association between diabetes-related exposures and AD incidence was modified by the presence of the apolipoprotein E (APOE) ε4 allele. These analyses have thus far produced inconsistent results, with some studies reporting a markedly greater risk of AD among diabetic ε4 carriers, relative to diabetic non-carriers [24], and others reporting no modification by APOE ε4 status [25]. Differences in risk by type of diabetes treatment have also been examined in a few cohorts. Treatment with insulin has been linked to greater AD/dementia risk, but this may simply be indicative of greater severity of diabetes [26, 27].

Exposure Modelling. Measurements of plasma glucose and insulin were obtained in a few studies, and were either modelled as continuous variables or in categories. The advantage of the first approach is that it considers AD risk across the full range of observed biomarker levels and potentially has greater statistical power if the relationship between the biomarker level and AD risk is log-linear. However, this approach can mask underlying relationships that are non-linear, such as the apparent association between insulin level and AD risk. The second approach, in which glucose or insulin levels are categorized, may capture non-linear relationships, but is dependent on the selection of cut-points which may differ between studies, and also assumes that AD risk is the same for all persons within the range of a specified exposure category. For glucose, the cut-points used were generally based on the diagnostic boundaries defined by established criteria. However, differences between the glucose measures used made comparison between the few available studies difficult. For plasma insulin, diagnostically defined categories do not exist, and selection of cut-points was based on quartiles or data-driven thresholds which varied across studies.

Results from Other Lines of Research

Numerous studies have evaluated diabetes mellitus in relation to cognitive decline among cognitively intact older adults, with findings that suggest greater declines among those with diabetes [28-30]. Diabetes has also been associated with a greater rate of conversion from mild cognitive impairment (MCI) to AD [31], and faster cognitive decline among patients with incident AD [32].

The few studies investigating the link between markers of pre-diabetes and cognitive decline have yielded mixed results. Recently, two studies among non-diabetics in well-established cohorts found no evidence of an association between cognitive decline and elevated fasting glucose [33] or HbA1c levels [34]. By contrast, in another prospective study, pre-diabetic women with impaired fasting glucose had greater cognitive decline than non-diabetic women [35]. Among non-diabetics, markers of higher insulin production and greater insulin resistance have generally correlated with steeper cognitive decline [36, 37], although this relationship has not been observed in some cohorts [33].

Multiple population-based studies have found greater degrees of cerebral atrophy on structural MRI among diabetic older adults relative to controls [38]. Further, there is also evidence indicating that progression of brain atrophy is accelerated among diabetics [39, 40]. Studies examining exposures other than diabetes diagnosis in relation to AD imaging biomarkers are just getting underway, but one cross-sectional study found increased insulin resistance to be associated with more severe hippocampal atrophy among non-diabetics [41].

The role of insulin in normal memory and its hypothesized importance in the pathogenesis of AD has spurred interest in the use of insulin therapy as a treatment for AD. Rosiglitazone, a diabetes medication that improves insulin sensitivity, failed to outperform placebo in a six-month randomized, double-blinded Phase III trial examining cognition and global functioning in persons with mild-to-moderate probable AD [42]. Rosiglitazone has also recently been linked to an increased risk of serious heart complications [43, 44]. Intranasal insulin administration was evaluated in a small, four-month randomized, placebo-controlled trial of 104 adults with amnestic MCI and AD and showed promising results for cognitive and functional outcome measures. Participants assigned to insulin treatment showed improvements in delayed memory and maintained their level of functioning over the trial period [45]. Trials of other diabetes medications and similar intranasal insulin interventions are ongoing.

The impact of diabetic control and treatment on cognitive outcomes has also been assessed in a few studies. Observational studies have given inconsistent results, with one study [29] linking lower HbA1c levels with better cognitive function, and another [34] finding no such link. A few clinical trials have also examined the effect of glucose-lowering treatments on cognitive outcomes. In a large trial of the impact of tighter control on type I diabetics (mean baseline age 27), assignment to the intensive versus the conventional treatment regimen was not associated with change in cognitive scores over 18 years of follow-up. However, elevated mean levels of HbA1c were associated with significantly greater decline [4]. Among type 2 diabetics, two large trials have found no effect of intensive versus standard glucose-lowering strategies on cognitive outcomes in populations aged 55 and older. In one of these trials [46], cognitive decline and dementia incidence were unaffected by glucose-lowering regimens. In the other trial [47], there was no difference over the 40-month follow-up between the two glucose-control regimens on any of the cognitive domains assessed. Participants assigned to the intensive-treatment regimen did have less brain atrophy over the follow-up period, but also had greater mortality rates, which led to the suspension of the intensive-control arm approximately 3 years into the trial.

Discussion and Recommendations

Overall, these data suggest a role for diabetes mellitus in the development of AD and cognitive decline, and add one more reason to bolster efforts to prevent diabetes through exercise, prevention of weight gain, and weight reduction [48]. Preliminary evidence suggests that the earlier stage of impaired glucose tolerance may also be linked to increased AD risk.

While some evidence suggests that AD risk is lower among persons with controlled diabetes, the optimal method of glucose control has not been established. Standard glucose control is critical in diabetes to prevent symptomatic hyperglycemia (and, among type 1 diabetes, ketoacidosis), as well as to lower the risk of microvascular and macrovascular complications. It is plausible that such regimens may also provide benefit for cognitive outcomes, either via direct effects on the brain, or by lowering cardiovascular disease risk. More intensive treatment regimens that aim to lower HbA1c levels below 7.0% reduce the occurrence of microvascular complications of diabetes, but have not been shown to provide any benefit over standard treatment regimens on cognitive outcomes [46, 47]. Tight glucose control also entails some risks, especially among older adults [49], so more research is required on the effects of such regimens on cognitive outcomes.


References

1. Breteler, M.M., Vascular involvement in cognitive decline and dementia. Epidemiologic evidence from the Rotterdam Study and the Rotterdam Scan Study. Ann N Y Acad Sci, 2000. 903: p. 457-65. Abstract

2. Stampfer, M.J., Cardiovascular disease and Alzheimer's disease: common links. J Intern Med, 2006. 260(3): p. 211-23. Abstract

3. Biessels, G.J., et al., Risk of dementia in diabetes mellitus: a systematic review. Lancet Neurol, 2006. 5(1): p. 64-74. Abstract

4. DCCT/EDIC Group, Long-term effect of diabetes and its treatment on cognitive function. N Engl J Med, 2007. 356(18): p. 1842-52.

5. Cholerton, B., L.D. Baker, and S. Craft, Insulin resistance and pathological brain ageing. Diabet Med, 2011. 28(12): p. 1463-75. Abstract

6. Power, M.C., et al., The association between blood pressure and incident Alzheimer disease: a systematic review and meta-analysis. Epidemiology, 2011. 22(5): p. 646-59. Abstract

7. Luchsinger, J.A., et al., Measures of adiposity and dementia risk in elderly persons. Arch Neurol, 2007. 64(3): p. 392-8. Abstract

8. Tabak, A.G., et al., Prediabetes: a high-risk state for diabetes development. Lancet, 2012. 379(9833): p. 2279-90.

9. WHO, Definition and diagnosis of diabetes mellitus and intermediate hyperglycaemia: report of a WHO/IDF consultation., 2006.

10. American Diabetes Association, Diagnosis and classification of diabetes mellitus. Diabetes Care, 2012. 35 Suppl 1: p. S64-71.

11. Faerch, K., et al., Pathophysiology and aetiology of impaired fasting glycaemia and impaired glucose tolerance: does it matter for prevention and treatment of type 2 diabetes? Diabetologia, 2009. 52(9): p. 1714-23.

12. Abdul-Ghani, M.A., D. Tripathy, and R.A. DeFronzo, Contributions of beta-cell dysfunction and insulin resistance to the pathogenesis of impaired glucose tolerance and impaired fasting glucose. Diabetes Care, 2006. 29(5): p. 1130-9.

13. Smith, N.L., et al., Fasting and 2-hour postchallenge serum glucose measures and risk of incident cardiovascular events in the elderly: the Cardiovascular Health Study. Arch Intern Med, 2002. 162(2): p. 209-16.

14. Ohara, T., et al., Glucose tolerance status and risk of dementia in the community: the Hisayama study. Neurology, 2011. 77(12): p. 1126-34. Abstract

15. Grundy, S.M., et al., Diagnosis and management of the metabolic syndrome: an American Heart Association/National Heart, Lung, and Blood Institute Scientific Statement. Circulation, 2005. 112(17): p. 2735-52.

16. Matthews, D.R., et al., Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia, 1985. 28(7): p. 412-9.

17. Wallace, T.M., J.C. Levy, and D.R. Matthews, Use and abuse of HOMA modeling. Diabetes Care, 2004. 27(6): p. 1487-95.

18. Expert Panel on Detection Evaluation and Treatment of High Blood Cholesterol in Adults, Executive Summary of The Third Report of The National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, And Treatment of High Blood Cholesterol In Adults (Adult Treatment Panel III). JAMA, 2001. 285(19): p. 2486-97.

19. Diabetes mellitus: a major risk factor for cardiovascular disease. A joint editorial statement by the American Diabetes Association; The National Heart, Lung, and Blood Institute; The Juvenile Diabetes Foundation International; The National Institute of Diabetes and Digestive and Kidney Diseases; and The American Heart Association. Circulation, 1999. 100(10): p. 1132-3.

20. Roglic, G., et al., The burden of mortality attributable to diabetes: realistic estimates for the year 2000. Diabetes Care, 2005. 28(9): p. 2130-5.

21. Weuve, J., et al., Accounting for bias due to selective attrition: the example of smoking and cognitive decline. Epidemiology, 2012. 23(1): p. 119-28.

22. Chatfield, M.D., C.E. Brayne, and F.E. Matthews, A systematic literature review of attrition between waves in longitudinal studies in the elderly shows a consistent pattern of dropout between differing studies. J Clin Epidemiol, 2005. 58(1): p. 13-9.

23. Euser, S.M., et al., Measuring cognitive function with age: the influence of selection by health and survival. Epidemiology, 2008. 19(3): p. 440-7.

24. Irie, F., et al., Enhanced risk for Alzheimer disease in persons with type 2 diabetes and APOE epsilon4: the Cardiovascular Health Study Cognition Study. Arch Neurol, 2008. 65(1): p. 89-93. Abstract

25. Akomolafe, A., et al., Diabetes mellitus and risk of developing Alzheimer disease: results from the Framingham Study. Arch Neurol, 2006. 63(11): p. 1551-5. Abstract

26. Ott, A., et al., Diabetes mellitus and the risk of dementia: The Rotterdam Study. Neurology, 1999. 53(9): p. 1937-1942. Abstract

27. Luchsinger, J.A., et al., Diabetes mellitus and risk of Alzheimer's disease and dementia with stroke in a multiethnic cohort. Am J Epidemiol, 2001. 154(7): p. 635-41. Abstract

28. Okereke, O.I., et al., Type 2 diabetes mellitus and cognitive decline in two large cohorts of community-dwelling older adults. J Am Geriatr Soc, 2008. 56(6): p. 1028-36. Abstract

29. Yaffe, K., et al., Diabetes, Glucose Control, and 9-Year Cognitive Decline Among Older Adults Without DementiaDiabetes and Risk of Cognitive Decline. Arch Neurol, 2012: p. 1-6.

30. Cukierman, T., H.C. Gerstein, and J.D. Williamson, Cognitive decline and dementia in diabetes--systematic overview of prospective observational studies. Diabetologia, 2005. 48(12): p. 2460-9. Abstract

31. Xu, W., et al., Accelerated progression from mild cognitive impairment to dementia in people with diabetes. Diabetes, 2010. 59(11): p. 2928-35.

32. Helzner, E.P., et al., Contribution of vascular risk factors to the progression in Alzheimer disease. Arch Neurol, 2009. 66(3): p. 343-348. Abstract

33. Euser, S.M., et al., A prospective analysis of elevated fasting glucose levels and cognitive function in older people: results from PROSPER and the Rotterdam Study. Diabetes, 2010. 59(7): p. 1601-7.

34. Christman, A.L., et al., Glycated haemoglobin and cognitive decline: the Atherosclerosis Risk in Communities (ARIC) study. Diabetologia, 2011. 54(7): p. 1645-52.

35. Yaffe, K., et al., Diabetes, impaired fasting glucose, and development of cognitive impairment in older women. Neurology, 2004. 63(4): p. 658-63.

36. Van Oijen, M., et al., Fasting insulin levels and cognitive decline in older women without diabetes. Neuroepidemiology, 2008. 30(3): p. 174-179.

37. Okereke, O.I., et al., Plasma C-peptide levels and rates of cognitive decline in older, community-dwelling women without diabetes. Psychoneuroendocrinology, 2008. 33(4): p. 455-61.

38. van Harten, B., et al., Brain imaging in patients with diabetes: a systematic review. Diabetes Care, 2006. 29(11): p. 2539-48.

39. van Elderen, S.G., et al., Progression of brain atrophy and cognitive decline in diabetes mellitus: a 3-year follow-up. Neurology, 2010. 75(11): p. 997-1002.

40. de Bresser, J., et al., Progression of cerebral atrophy and white matter hyperintensities in patients with type 2 diabetes. Diabetes Care, 2010. 33(6): p. 1309-14.

41. den Heijer, T., et al., Type 2 diabetes and atrophy of medial temporal lobe structures on brain MRI. Diabetologia, 2003. 46(12): p. 1604-10. Abstract

42. Gold, M., et al., Rosiglitazone monotherapy in mild-to-moderate Alzheimer's disease: results from a randomized, double-blind, placebo-controlled phase III study. Dement Geriatr Cogn Disord, 2010. 30(2): p. 131-46.

43. Loke, Y.K., C.S. Kwok, and S. Singh, Comparative cardiovascular effects of thiazolidinediones: systematic review and meta-analysis of observational studies. BMJ, 2011. 342: p. d1309.

44. Nissen, S.E. and K. Wolski, Effect of rosiglitazone on the risk of myocardial infarction and death from cardiovascular causes. N Engl J Med, 2007. 356(24): p. 2457-71. Abstract

45. Craft, S., et al., Intranasal insulin therapy for Alzheimer disease and amnestic mild cognitive impairment: a pilot clinical trial. Arch Neurol, 2012. 69(1): p. 29-38. Abstract

46. ADVANCE Collaborative Group, Intensive blood glucose control and vascular outcomes in patients with type 2 diabetes. N Engl J Med, 2008. 358(24): p. 2560-72.

47. Launer, L.J., et al., Effects of intensive glucose lowering on brain structure and function in people with type 2 diabetes (ACCORD MIND): a randomised open-label substudy. Lancet Neurol, 2011. 10(11): p. 969-77. Abstract

48. Gillies, C.L., et al., Pharmacological and lifestyle interventions to prevent or delay type 2 diabetes in people with impaired glucose tolerance: systematic review and meta-analysis. BMJ, 2007. 334(7588): p. 299.

49. Skyler, J.S., et al., Intensive glycemic control and the prevention of cardiovascular events: implications of the ACCORD, ADVANCE, and VA Diabetes Trials: a position statement of the American Diabetes Association and a Scientific Statement of the American College of Cardiology Foundation and the American Heart Association. J Am Coll Cardiol, 2009. 53(3): p. 298-304.