Intranasal Insulin and Post-stroke Cognition: A Pilot Study

Almost two-thirds of survivors have cognitive impairment (CI), manifested as memory, language, and judgement problems. Post-stroke CI at 2 weeks is a significant predictor of long-term functional outcome, and more generally, cognitive impairments have a major impact on functional outcome and ability to participate in rehabilitation. CI is associated with increased systemic inflammation. Intranasally-administered insulin is a promising new therapy for enhancing memory in patients with mild cognitive impairment (MCI) and Alzheimer's disease (AD), shown in multiple randomized controlled studies. Likely mechanisms of benefit are intranasal insulin's ability to restore normal cerebral insulin signaling. Based on the overlap in cerebral insulin resistance that occurs in both AD and post-stroke CI, we have designed an innovative proof-of-concept, feasibility trial designed to provide pilot data as to whether post-stroke survivor CI and caregiver burden is improved with intranasal insulin early after stroke. We will explore the impact of intranasal insulin on inflammatory biomarkers, since inflammation is a major underlying cause of CI, as shown by others and in our preliminary studies of VCAM-1. Specific Aims are: 1. Determine if patients with ischemic stroke randomized to intranasal insulin 20 IU BID for 3 weeks have improved cognition, compared to patients who receive intranasal saline. Primary outcome is a composite of (a) memory and executive function z scores. 2. To assess the impact of intranasal insulin vs saline on change in inflammatory biomarker levels (VCAM-1, TNF-alpha, TNFR-I and II) before and after the treatment period. 3. To measure differences in burden among caregivers of participants in the intranasal insulin vs intranasal saline groups. We will prospectively randomize 40 subjects to intranasal insulin (40 IU) vs saline treatment. Following baseline cognitive testing 2 weeks post stroke, subjects will receive the assigned treatment for 3 weeks, followed by a 3-week washout period, with cognitive testing performed after the treatment and washout periods and again at 20 weeks. The proposed study will provide data on a promising method for treating cognitive function in stroke patients. If effective, our pilot data will set the stage for larger phase III clinical trials.

Hypothesis: Intranasal insulin will improve post-stroke cognitive impairment if administered in the early post-acute stroke phase compared with placebo. Design: Block randomized, double-blind placebo-controlled trial of 40 patients, 20 per group. Study population: Ischemic stroke patients presenting within 48 hours of symptom onset and without moderate to severe aphasia or homonymous hemianopia who have evidence of cognitive impairment on the Montreal Cognitive Assessment (MoCA). Recruitment: Patients identified during the acute hospitalization for ischemic stroke, inpatient rehabilitation, or in the post-stroke follow-up clinic. Patients from other hospitals and clinics are also eligible. Eligibility will be assessed through screening MoCA and if in the target range, will be invited to participate in the study. Consent: Patients will sign informed consent if eligible. Randomization and Blinding: The participant will be randomized and undergo the full baseline cognitive battery. Caregivers will also sign a separate consent to complete the Caregiver Strain Index. Subjects will be randomized to intranasal insulin (20 IU BID) vs saline BID according to a stratified permuted block randomization of 4 as per our statistician, stratified by age 40 to 69 years vs > 70 years) and presence of language deficit (based on NIHSS) at the time of randomization. Insulin and saline will be packaged in identical single-dose ampules that will be opened and inserted into the chamber of the VianaseTM device. Ampoules will be dispensed in 3 week supplies at each study visit. Package labels will instruct participants to administer each dose 30-60 minutes after breakfast and dinner. If a dose is missed, it will not be replaced. The insulin and saline will be identical in packaging except for randomization code. Patients/caregivers, investigators, and outcome assessors will be masked to the group assignment. All subjects will receive their first dose in the clinic and wait for 2 hours to determine any adverse effects of the inhaled dose, while remaining blinded to the treatment. The glucose level will be measured with the glucometer to monitor for hypoglycemia and documented. All subjects will additionally check peak dose blood sugar levels with a glucometer, 3 times per week during insulin or saline treatment. The patients and caregivers will receive a handbook which describes the method for loading the device, inhalation, methods for fingersticks, use of the glucometer and test strips, a blood sugar log, and the signs and symptoms of hypoglycemia. They will also document any other adverse effects on the weekly log. Procedures: Cognitive impairment measured early after stroke predicts functional outcomes in stroke patients at 13 months. Furthermore, waiting until 6 weeks after the stroke to conduct baseline cognitive testing and treatment allows enough recovery and completion of intensive rehabilitation for those patients who may not be able to complete cognitive testing immediately after stroke, but takes advantage of the potential benefits of insulin in the early phase, shown in animal studies. In addition, we potentially reduce the variability of the biomarkers by avoiding the acute phase of stroke. The frequency of cognitive testing is based on trials of intranasal insulin for MCI and AD. Patients with a history of psychiatric medication use for the management of mental and emotional disorders prior to their stroke will not be excluded from the study. However, as these medications may have an effect on cognitive testing, initiation of psychiatric medications should be avoided from the time of study enrollment through the post wash out period (9 weeks total) unless it is clearly indicated to be in the patient's best interest. Patients will be informed of the prohibited medications during consent and will be asked to immediately inform the study team of any new medications or changes in existing medications. Baseline Data collection. We will collect the baseline data described below as in the VCI Harmonization standards: Demographics: Birth date, sex, race/ethnicity, years in current country of residence, number of years of education, occupation, literacy, living situation and level of independence, type of residence, marital status, and name and contact information for a family member or caregiver will be collected. Proxy/informant information collected: Birth date, sex, race/ethnicity, relationship and length of time of relationship with the patient, education, and living status with respect to the subject. Family history information collected: First-degree relatives with stroke, myocardial infarction, and dementia, approximate age at onset, and age at death. Stroke hospital data collected: Severity (NIHSS) at onset, location of stroke, vascular territory, modified Rankin score at discharge, short physical performance battery score, length of stay, stroke complications, stroke prevention medications at discharge, stroke risk factors (diabetes, hypertension, hyperlipidemia, smoking, alcohol/substance abuse, atrial fibrillation, carotid stenosis, hypercoagulable state), and stroke subtype (NINDS classification). Intranasal insulin administration: We will use an innovative investigational device developed by Kurve Technology (Fig 5A). Typical spray bottle administration results in large droplets that penetrate only within the first 20% of the lower nasal cavity, and due to gravity and insufficient airflow, ~90% of the droplets wind up in the stomach (Fig.5B). The ViaNaseTM device delivers a substance throughout the nasal cavity, to the olfactory region and paranasal sinuses, thereby maximizing access to nose-to-brain channels (Fig.5C). This distribution occurs because droplet size is adjusted according to the weight of the substance, through an individually optimized droplet generator resulting in maximal vortical distribution (Fig.5D). We have used the ViaNaseTM device with excellent results in two pilot trials (described above).12,13 A total volume of 2 mL of insulin or placebo (saline) will be administered each time. Caregivers will supervise participants in administering intranasal treatment 2 times per day, after breakfast and dinner. Participants and caregivers will be trained in use of the delivery device. In previous studies, participants with aMCI/AD have found the device to be easy and pleasant to use, with compliance rates of 95 to 97%. Cognitive Outcome Analysis: The analysis will be based on intention to treat. The primary outcomes are analyzed via composite z scores at 3 weeks of treatment in both groups. The summary scores of the SIS, IADL scale and SIS-16 will be compared from baseline to 3 weeks post treatment in both groups. Functional covariates: Disability and physical function measures will be collected to understand the trajectory of motor recovery and will be mapped to the cognitive trajectory from baseline (2 weeks to 20 weeks post-stroke). Exploratory analyses will be performed to determine whether INI leads to an improvement in IADLs, SIS-16, and mRS. Statistical Analysis: The 3 week intervention effect of INI on the cognitive outcomes will be assessed using analysis of covariance (ANCOVA) adjusting for age, language deficit, and baseline cognitive score. We will assume a Type I error rate alpha = 0.05 for all analyses. Adjusting for baseline characteristics will help attenuate the potential effect of differential dropout, but we will conduct a sensitivity analysis using multiple imputation to ensure dropout does not unduly affect our results. The observed estimates, standard deviations, and dropout rates will be used for the design of future studies of INI treatment for stroke patients. Exploratory analyses will examine response by subgroup for evidence of an enhanced treatment effect, including subgroups defined by participant age (<70 vs. >70 years), baseline cognitive impairment and the presence or absence of any degree of language deficit (aphasia). Covariates will be utilized so that they do not overlap with clinical subgroups. Because of the relatively small sample size, if the prespecified subgroups based on continuous characteristics are not sufficiently large, new clinically meaningful subgroups may be formed for exploratory analyses. Power Analysis: The primary purpose of the pilot study is to demonstrate feasibility and safety. The proposed study will collect data to estimate power in future trials, identify instruments most sensitive to the proposed cognitive effects, and anticipate recruitment and retention rates in the target population. However, the proposed study will have 83% power to detect a 1 standard deviation difference in cognitive scores between groups at 3 weeks (i.e. effect size = 1.0), assuming normally distributed outcomes, alpha = 0.05, and a 10% dropout rate.

Intranasal Insulin (20 IU BID): Humulin insulin packaged is in single-dose ampules and inserted into the VianaseTM chamber. Ampoules are dispensed in 3 week supplies per study visit. Patients/caregivers, investigators, and outcome assessors are masked to group assignment. Subjects will receive their first dose in clinic, and wait 2 hrs to determine any adverse effects of inhaled dose. Glucose levels will be measured to monitor for hypoglycemia and documented. All subjects will check peak dose blood sugar levels with glucometer, 3 times per week during insulin treatment.

Intranasal saline: Saline is packaged in single-dose ampules and inserted into the VianaseTM chamber. Ampoules are dispensed in 3 week supplies per study visit. Patients/caregivers, investigators, and outcome assessors are masked to group assignment. Subjects will receive their first dose in clinic, and wait 2 hrs to determine any adverse effects of inhaled dose. Glucose levels will be measured to monitor for hypoglycemia and documented. All subjects will check peak dose blood sugar levels with glucometer, 3 times per week during insulin treatment.

Hopkins Verbal Learning Test-Revised (HVLT-R) verbal learning and memory and available to facilitate repeat administration in future testing (10 minutes with delay and recognition). The delayed recall is used as the outcome in this study. Brief Visual Memory Test-Revised (BVMT-R) is a measure of nonverbal learning and memory captured with immediate and delayed free recall trials, and a recognition memory task (10 minutes with delay and recognition). The retained score normalized by age, is used as the outcome for this study. The t scores for both HVLT-R delayed recall and BVMT-R retained are normalized from the raw scores based on age. The t score ranges from 0-100 with 0 denoting a worse outcome. The Composite is the average of the mean t scores for each component.

Hopkins Verbal Learning Test-Revised (HVLT-R) verbal learning and memory and available to facilitate repeat administration in future testing (10 minutes with delay and recognition). The delayed recall is used as the outcome in this study. Brief Visual Memory Test-Revised (BVMT-R) is a measure of nonverbal learning and memory captured with immediate and delayed free recall trials, and a recognition memory task (10 minutes with delay and recognition). The retained score is used as the outcome for this study. The t scores for both HVLT-R delayed recall and BVMT-R retained are normalized from the raw scores based on age. The t score ranges from 0-100 with 0 denoting a worse outcome. The Composite is the average of the mean t scores for each component.

Hopkins Verbal Learning Test-Revised (HVLT-R) verbal learning and memory and available to facilitate repeat administration in future testing (10 minutes with delay and recognition). The delayed recall is used as the outcome in this study. Brief Visual Memory Test-Revised (BVMT-R) is a measure of nonverbal learning and memory captured with immediate and delayed free recall trials, and a recognition memory task (10 minutes with delay and recognition). The retained score is used as the outcome for this study. The t scores for both HVLT-R delayed recall and BVMT-R retained are normalized from the raw scores based on age. The t score ranges from 0-100 with 0 denoting a worse outcome. The Composite is the average of the mean t scores for each component.

The Trail Making test-B is a test of sequencing between numbers and letters, and the score is the amount of time needed to complete the test. The t score is the raw score for time to completion and errors, normalized by age and education, and ranges from 0 (poor performance) to 100 (good performance). WAIS Digit Span subtest Reverse is a measure of working memory, and requires repetition of increasingly longer strings of digits in the reverse order. The t scores are raw scores normalized by age and range from 0 to 100. WAIS-III Digit-Symbol Coding is a measure of visuomotor processing speed, and requires involves rapidly coding geometric symbols given a number, by using a legend of number-symbol pairs at the top of the page. T scores are raw scores normalized by age, and range from 0 to 100. The composite outcome included average of the mean t scores for Trailmaking B, Digit span reverse, and digit-symbol coding.

The Trail Making test-B is a test of sequencing between numbers and letters, and the score is the amount of time needed to complete the test. The t score is the raw score for time to completion and errors, normalized by age and education, and ranges from 0 (poor performance) to 100 (good performance). WAIS Digit Span subtest Reverse is a measure of working memory, and requires repetition of increasingly longer strings of digits in the reverse order. The t scores are raw scores normalized by age and range from 0 to 100. WAIS-III Digit-Symbol Coding is a measure of visuomotor processing speed, and requires involves rapidly coding geometric symbols given a number, by using a legend of number-symbol pairs at the top of the page. T scores are raw scores normalized by age, and range from 0 to 100. The composite outcome included average of the mean t scores for Trailmaking B, Digit span reverse, and digit-symbol coding.

The Trail Making test-B is a test of sequencing between numbers and letters, and the score is the amount of time needed to complete the test. The t score is the raw score for time to completion and errors, normalized by age and education, and ranges from 0 (poor performance) to 100 (good performance). WAIS Digit Span subtest Reverse is a measure of working memory, and requires repetition of increasingly longer strings of digits in the reverse order. The t scores are raw scores normalized by age and range from 0 to 100. WAIS-III Digit-Symbol Coding is a measure of visuomotor processing speed, and requires involves rapidly coding geometric symbols given a number, by using a legend of number-symbol pairs at the top of the page. T scores are raw scores normalized by age, and range from 0 to 100. The composite outcome included average of the mean t scores for Trailmaking B, Digit span reverse, and digit-symbol coding.

These tasks of verbal fluency provide the subject one minute to say as many words as possible. For Animal Naming, any living creature is counted. For FAS, words beginning with a given letter (F, A, and then S) are counted excluding proper nouns, numbers, and variations of the same word. Raw scores are converted to t scores, normalized by age, and range from 0 (worse) to 100 (better) outcome.

These tasks of verbal fluency provide the subject one minute to say as many words as possible. For Animal Naming, any living creature is counted. For FAS, words beginning with a given letter (F, A, and then S) are counted excluding proper nouns, numbers, and variations of the same word. Raw scores are converted to t scores, normalized by age, and range from 0 (worse) to 100 (better) outcome.

These tasks of verbal fluency provide the subject one minute to say as many words as possible. For Animal Naming, any living creature is counted. For FAS, words beginning with a given letter (F, A, and then S) are counted excluding proper nouns, numbers, and variations of the same word. Raw scores are converted to t scores, normalized by age, and range from 0 (worse) to 100 (better) outcome.

The MoCA is a cognitive screening instrument which provides information about overall mental status. It is useful for identifying overall cognitive impairment, and also provides information about core cognitive domains, such as visuospatial abilities, attention, executive function, language, orientation, and memory. Score ranges from 0-30 with scores of 26-30 denoting no cognitive impairment, 21-25 mild cognitive impairment and 20 or lower, dementia or severe cognitive impairment.

The PHQ-9 is a measure of depression and is comprised of 9 questions with likert scale responses related to how frequent depressive symptoms are occurring. The likert score for each question is totaled for the final score, which ranges from 0-27 with a score of 27 denoting a higher severity of depression.

A narrative of 44 informational bits is read and the recalled information is recorded immediately and after 20 minutes. Information retained over the delay are calculated as delayed recall/ immediate recall. The participant is asked to recite it immediately, and then following a 20-minute delay. Delayed score (the outcome analyzed here) ranges from 0 to 44 with a higher score denoting a better ability to recall information.

Lawton Instrumental ADL scale includes 8 domains related to higher level function, including using a telephone, handling money, ability to take medications correctly, independence with transportation, food preparation, shopping, laundry, and housekeeping. Score ranges from 0-8 with 0 denoting no independence with IADLs, and 8 the highest level of independence with IADLs.

13 questions related to caregiver strain, administered to caregivers separately from the stroke survivor. Score ranges from 0-13 with a score of 7 or more indicating significant caregiver strain.

The mRS Score is a score for disability after stroke. A score of 0 indicates no symptoms remaining and normal pre-stroke activity, 1 indicates no disability despite symptoms, 2 indicates mild disability but still independent, 3 indicates moderate disability but able to walk, 4 indicates moderate to severe disability and requiring significant help from others, 5 indicates 24 hour nursing care and complete dependence on others, and 6 is death.

Inclusion Criteria: Ischemic stroke and measurable deficit on the initial NIHSS (> 1) Cognitive impairment within the 5th and 50th percentiles for age, race, and education based on Montreal Cognitive Assessment (MoCA) or 2 out of 5 delayed recall or less on the MoCA. Able to sign informed consent, have a caregiver, and live within a reasonable driving distance from Wake Forest Baptist Medical Center. Exclusion Criteria: Patients under age 40 or 90 years or older Living in skilled nursing facility Severe stroke deficits at 4 weeks that prohibit participation in cognitive testing (global or receptive aphasia, or severe expressive aphasia) Diabetes requiring insulin Psychiatric disorders Severe head trauma Alcoholism Neurologic disorders other than stroke Renal disease hepatic disease chronic obstructive pulmonary disease unstable cardiac disease those with prior deficits in ADLs and IADLs