The study will examine the effects of intranasally administered long-acting insulin detemir on cognition in persons with Alzheimer's disease (AD) or amnestic mild cognitive impairment (aMCI). The rationale for these studies is derived from growing evidence that insulin contributes to multiple brain functions, and that insulin dysregulation can contribute to AD pathogenesis. Thus, therapies aimed at restoring normal insulin signaling in the CNS may have beneficial effects on brain function. Intranasal administration of insulin increases insulin signaling in the brain without raising peripheral levels and causing hypoglycemia. Insulin detemir is an insulin analogue that may have better action in brain than other insulin formulations because of its albumin binding properties. The investigators will test the therapeutic effects of intranasally-administered insulin detemir in a study in which participants will receive insulin detemir, regular insulin, or placebo over a four month period. The investigators will test the hypothesis that insulin and insulin detemir will both improve memory and daily functioning in persons with AD/aMCI compared with placebo, but that insulin detemir will have the greatest effect.

It is well-known that insulin, a hormone that is naturally secreted by the pancreas, plays an important physiological role by regulating blood sugar levels in the body. Researchers now know that insulin plays many important roles in the brain as well. Insulin seems to be especially active in the part of the brain that corresponds to learning and memory. Studies have shown that when people have insufficient insulin in the brain (which, for example, is the case with Type-II diabetes), they are increasingly at risk to develop memory problems and Alzheimer's disease. In a past study, the investigators administered intravenous insulin to participants and found that it improves memory. However, that particular method would not be a practical intervention for people with Alzheimer's disease due to the risks of hypoglycemia or exacerbation of insulin resistance. Instead, the investigators use an "intranasal" method of administration, in which the insulin is inserted into a device, and administered intranasally. In this method, the insulin travels directly to the brain, and bypasses the body. Past studies have also demonstrated that this can be a reliable way to improve memory, and it does not change the body's blood glucose levels.

In our past studies, investigators have used regular insulin, which lasts about 3-4 hours and creates a similar "spike" in insulin that one would have after eating a meal. However, in normal physiology, the pancreas also releases small and more constant "pulses" of insulin throughout the day and night, establishing a base level of insulin. Accordingly, several longer-lasting types of insulin are now available that last closer to 10-12 hours, mimicking that basal level of insulin. The current study uses a long-lasting type of insulin called "insulin detemir," to determine if learning and memory will benefit from a more consistent supplement of insulin. The investigators want to determine whether this treatment can benefit people who already have a memory impairment-either they have a diagnosis of Alzheimer's disease (AD) or have a mild cognitive impairment (MCI), a condition that precedes Alzheimer's disease. The investigators will examine cognition, daily function, cerebral blood flow, and different markers of Alzheimer's disease that are in the blood and cerebral spinal fluid (CSF) as outcome measures.

The investigators have these specific aims:

1. We will test the hypothesis that compared to placebo, four months of treatment with intranasal insulin or insulin detemir will improve cognition and function in adults with AD or MCI, but that greater effects will be observed for insulin detemir.
2. We will examine the effects of intranasal insulin and insulin detemir on cerebral blood flow in adults with AD or MCI.
3. We will examine the effects of intranasal insulin and insulin detemir on CSF Aβ, tau and inflammatory markers in adults with AD or MCI.

To examine these hypotheses, the investigators are recruiting approximately 90 participants who have been diagnosed with AD or mild cognitive impairment. They will be randomly selected to take a placebo (saline), insulin detemir, or insulin. Cognition, the level of daily functioning, glucose tolerance, and cerebral blood flow will be tested before they begin the study drug, and after 16 weeks of the study drug. Some participants will also undergo a lumbar puncture both before beginning study drug and after 16 weeks of taking the study drug.

Statistical analysis will follow an intent-to-treat (ITT) approach; that is, subjects will be analyzed in their original randomized group regardless of adherence to group assignment. A completer analysis will also be performed, including only those subjects who successfully complete the treatment phase. Missing data will be handled using multiple imputation linear regression. We will conduct secondary analyses on other measures of cognition, daily function, cerebral blood flow, and CSF biomarkers. For ASL-MRI, following coregistration and processing, parametric maps will be generated to determine regional CBF values by treatment group. Secondary analyses will also examine treatment duration (2-month vs. 4-month) for all relevant outcomes. All models will be adjusted for age and an index of peripheral insulin sensitivity (derived from 120-minute OGTT glucose and insulin values) if statistically warranted, and posthoc contrasts will be performed when appropriate. Secondary analyses will also evaluate whether treatment response of cognition, daily function, CSF and plasma markers, and insulin differ according to APOE4 genotype. Although these analyses will be exploratory due to possible limited APOE4 by treatment arm cell size, the data will be examined for statistical trends that warrant further exploration in larger trials. Other secondary analyses will examine associations among treatment-related outcomes using scores derived from multiple regression of data collected during the treatment phase residualized with respect to baseline values.

Inclusion Criteria:

• Age 50-89
• Diagnosed with mild cognitive impairment, or mild/moderate AD

Exclusion Criteria:

• Excessively high or low blood pressure, heart rate
• Pre-existing diabetes not controlled by exercise/diet
• Previous/current use of insulin
• Significant elevations in lipids, liver enzymes
• Menstrual period within the last 12 months
• Significant neurological or medical disorder (other than AD)
• Significant use of nasal decongestants
• Current use of anti-psychotic, anti-convulsive, anxiolytic, glucocorticoids, or sedative medications

There is no plan to share individual participant data with other researchers.