Effect of intranasally administered insulin on cerebral blood flow and perfusion; a randomized experiment in young and older adults

Plasma glucose and insulin trajectories

We measured the peripheral (venous) glucose and insulin trajectories throughout an experimental period of 90 minutes, which comprised measurements before, during and after MRI scanning. As shown in Figure 1, the trajectories of blood glucose and insulin were similar and stable in insulin compared to placebo conditions throughout the experimental period, both in young and older adults. In the young adults, the mean (SD) glucose and insulin levels over the 90- minute period were 4.57 (0.3) mmol/L and 5.25 (2.7) mU/L respectively under placebo condition, while the mean (SD) glucose and insulin levels were 4.61 (0.5) mmol/L and 5.71 (2.6) mU/L respectively under insulin condition. Likewise, in the older adults, the mean (SD) glucose and insulin levels over the 90- minute period were 4.87 (0.5) mmol/L and 5.75 (3.4) mU/L respectively under placebo condition, while the mean (SD) glucose and insulin levels were 4.87 (0.5) mmol/L and 5.30 (3.6) mU/L respectively under insulin condition.

Quantitative flow

The mean flow through the main cerebropetal arteries supplying the brain after intranasal application of insulin compared to placebo in both the older and younger adults is presented in Table 2. Intranasal application of insulin did not significantly change mean blood flow through the cerebropetal arteries in either older adults nor in the young. The mean difference in total quantitative flow under insulin minus placebo conditions in the younger adults was 0.35 ml/sec (SD 2.59), while that of the older adults was -0.24 ml/sec (SD 1.66).

The variance (standard deviations) around the mean flow were consistently lower after insulin administration. A Levene’s test verified the inequality of the variances in total quantitative flow after intranasal application of insulin compared to placebo when data from young and old adults were combined (P=0.047). Similar trends towards lower variances in cerebral blood flow after intranasal insulin application were observed in young and older adults.

ASL perfusion

Perfusion through specific cortical regions of the gray matter is shown in Table 3. In addition to cortical regions, we also assessed perfusion in the thalamus and the hypothalamus. In younger adults, perfusion through the whole brain gray matter, frontal, parietal, temporal and occipital gray matter regions after intranasal administration of insulin was not significantly different when compared to placebo. In younger adults, intranasal administration of insulin did not change perfusion in the thalamus or in the hypothalamus.

In older adults, intranasal administration of insulin significantly increased perfusion through the occipital gray matter with 6.5% when compared to the administration of placebo (Table 3, P=0.001). Perfusion through the parietal gray matter was 4.3% increased after intranasal administration of insulin (P=0.034). Perfusion in the frontal and the temporal gray matter regions was not significantly changed after intranasal administration of insulin (Table 3). In older adults, intranasal administration of insulin significantly increased perfusion in the thalamus (P=0.003), but no significant changes were found for perfusion in the hypothalamus. After application of Bonferroni corrected significance threshold of P<0.007, only the changes in occipital grey matter and the thalamus remained significant. The change in perfusion after intranasal application of insulin for one representative participant is visualized in the gray matter perfusion map shown in Figure 2.

In addition to the whole gray matter regions, we also investigated the perfusion through the right and left hemisphere of each region of interest (ROI) separately. The directions of observed effects in the left/right ROIs were comparable to those observed for the whole lobe ROIs. For example, in older participants perfusion in the right parietal region under placebo compared to insulin condition was 65.58 ± 11.70 compared to 68.31 ± 11.17 (p=0.061); and perfusion in the left parietal region under placebo compared to insulin condition was 67.12 ± 10.41compared to 70.12 ± 12.71 (p=0.221). Similarly, in older participants perfusion in the right occipital region under placebo compared to insulin condition was 60.92 ± 11.07 compared to 65.15 ± 10.97 (p=0.006); and perfusion in the left occipital region under placebo compared to insulin condition was 61.49 ± 9.35 compared to 65.13 ± 11.61 (p=0.005). Likewise, in older participants perfusion in the right thalamus under placebo compared to insulin condition was 63.79 ± 6.72 compared to 68.52 ± 6.14 (p=0.008); and perfusion in the left thalamus under placebo compared to insulin condition was 62.84 ± 7.11 compared to 68.03 ± 7.63 (p=0.005).

Analyses were repeated after exclusion of two older subjects that were hypertensive. The results did not materially change after exclusion of the two hypertensive older subjects.

Figure 3 shows the paired gray matter perfusion in the parietal and occipital lobes and in the thalamus for every individual subject separately to illustrate individual differences between the placebo and insulin.

Ethics statement

This study titled “Maintaining health in old age through homeostasis: Switchbox Phase II” was approved by the Medical Ethical Committee of Leiden University Medical Center under protocol P13.164 and the Dutch competent authority (Centrale Commissie Mensengebonden Onderzoek (CCMO)) with protocol code number NL45043.058.13. The study is registered in the European Clinical Trials Database under number 2012-005650-29. All investigations have been conducted according to the principles expressed in the Declaration of Helsinki. All participants provided written informed consent after written and verbal description of the study was given.

Trial design and participants

The design is a double-blinded, randomized, cross-over, placebo controlled trial in volunteers from the general population. This study was aimed at examining the effects of intranasal insulin on circulatory metabolic and endocrine parameters, as well as on resting and functional brain activities using MRI, in older and young adults. Fourteen elderly and nine young relatively healthy men (age criteria 60-85 years for older adults and 18-35 years for young adults), with a body mass index (BMI) between 20 kg/m2 and 27 kg/m² were recruited from the general population. Exclusion criteria included fasting plasma glucose above 7 mmol/l, anemia (haemoglobin < 7.1 mmol/l), presence of anatomic deviations of the nose, any significant endocrine, neurological and cardiovascular diseases, or use of medication known to influence lipolysis, thyroid function, glucose metabolism, GH/IGF-1 secretion or any other hormonal axis. Furthermore, persons with a current smoking or alcohol addiction or history of substance abuse were excluded. Of the 23 participants that were contacted via telephone, three (older) participants were excluded after screening. The reasons for exclusion were uncontrolled high blood pressure (n=1) and renal insufficiency (n=2). Of these, all participants except for one (young) successfully completed the study. The reason for the drop-out was phobia for cannulas/ blood withdrawal.

Randomization, masking and intervention

Randomization was carried out by the pharmaceutical trial coordinator, while the researchers were blinded to what the participants received. Concealment of treatment allocation was ensured by delivery of trial medication in a standard plain vial (same for all treatment groups) that had been prepared and re-labelled by the hospital pharmacy. De-blinding was done at the end of the study.

The study interventions consisted of intranasal application of 40 IU insulin (Actrapid; Novo Nordisk, Mainz, Germany) or placebo (Sterile saline), delivered using the ViaNase Electronic Atomizer (Kurve Technology inc.). The ViaNase device had been specifically designed to deliver drugs to the olfactory region to maximize drug transport to the central nervous system. The device released a metered insulin (40 IU) or saline dose directly into the subject’s nostrils. A total volume of 0.4 mL (0.2 ml per nostril) was administered, which was inhaled by breathing evenly over a 2-minute period. In both right and left nostril, two doses each were delivered resulting in a total insulin dose of 40 IU. This method allowed administration of smaller particle sizes to increase drug deposition in the upper nasal cavity without transporting the drug to the lungs [18].

For the current study, results from two study visits, spaced apart by a week, on one of which they received insulin and on the other placebo, in a randomized manner were compared. All participants underwent exactly the same protocol during both study visits.

Experimental protocol

In the morning after a 10-hour overnight fast, participants arrived at the MRI facilities at 08.00 in the morning. A catheter was placed in a vein of the forearm of the non-dominant hand. First, a baseline venous blood sample was collected in a serum-separator (SST)-tube. Thereafter, every 10 minutes, 4 ml of blood was collected into a SST-tube and 2 ml into K3-EDTA tube. Ten minutes after the first blood withdrawal, the trial medication (placebo or 40 IU of insulin) was administered intranasally, under the supervision of the experimenter, with a ViaNase Electronic Atomizer. Participants were asked if they noticed a specific smell or noticed something about the applied treatment. However, participants were not informed that the smell could be associated with the treatment and therefore would remain unaware of the applied intervention. We checked this by questioning the participants of the awareness of the treatment condition and this did not coincide with the actual applied treatment after de-blinding of the study. About 20 minutes after the intranasal application, participants were transferred into the MRI for the scanning procedure (Figure 4). About 30 min after the intranasal application, the CBF measurement was performed in the MRI scanner. This time interval was chosen because previous studies have shown that it takes 30 minutes for insulin to reach maximal concentrations in the CSF [19], and a similar time window has been used previously in other studies where brain – related effects of intranasal insulin were studied [20–22].

Anthropometrics

During the first study day, height, weight, percentage of body fat, and waist and hip circumference of participants were measured. Weight (in kilograms) was divided by the squared height (in meters) to calculate the BMI.

Blood sampling and chemical analysis of blood samples

After each withdrawal, the K3-EDTA tubes were immediately placed on ice before centrifugation, whereas the serum tubes were kept at room temperature and centrifuged when the samples were clotted, usually between 30–60 minutes. Samples were centrifuged at 3520 RPM at 4 °C for 10 minutes. The EDTA plasma and serum samples were stored in two aliquots of 500 µl during the rest of the sampling at − 20 °C. After the sampling they were transferred to a − 80 °C freezer and stored until analysis.

All laboratory measurements were performed with fully automated equipment and diagnostics from Roche Diagnostics (Almere, The Netherlands). Glucose levels were measured using Hitachi Modular P800 from Roche (Almere, the Netherlands), with coefficient of variation (CV) for measurement less than 1%. Insulin levels were measured using the Immulite 2500 from DPC (Los Angeles, CA). CV was less than 6%.

Brain MRI protocol

The perfusion measurement reported in the current study was part of a longer scanning protocol (48 minutes in young participants and 24 minutes in older participants) in which we aimed to successively assess several MRI endpoints. Scans were performed on a whole body magnetic resonance system with a 3 Tesla field strength (Philips Medical Systems, Best, the Netherlands). 3D T1-weighted images were acquired with the following imaging parameters: echo time (TE) 4.6ms, repetition time (TR) 9ms, Flip 8°, field of view (FOV) = 224 × 177 × 168mm, scan duration ~5minutes. The pseudo Continuous Arterial Spin Labeling (pCASL) with TE/TR/Flip: 14ms/4.0s/90°, FOV 240 x 133 x 240mm, matrix 80 x 80mm, slices 19, labeling duration of 1650 ms, post labeling delay ranging from 1525 ms for the most inferior slice to 2155 ms for the most superior slice scan, total scan duration 6 minutes. The M0-scan with TE/TR/flip: 14ms/6.0s/90, FOV 240x133x240, Matrix 80 80, slices 19, scan duration 30 seconds. The pCASL labelling slice was positioned perpendicular to the common carotid arteries right below the bifurcation point of common carotid into the internal and external carotid artery. The phase-contrast quantitative flow (QF) scan was planned using 2 localizer angiograms in the sagittal and coronal planes and acquired with the following parameters: TR = 11ms; TE 7.5ms; flip angle = 10°; slice thickness = 5mm; field of view 150x103mm; voxel size 1.17x1.17mm; velocity sensitivity = 200cm/s, 20 signal averages. Time allocated to measure perfusion was 6 minutes.

Phase contrast MR Angiography

Quantitative flow measurements were performed using ungated phase-contrast MRA, based on 2 localizer MR angiograms in the sagittal and coronal orientation as shown in Figure 5. A region of interest (ROI) was drawn around the vessel lumen on the magnitude images by an experienced rater using Philips Software on a PACS (Philips Medical Systems, Best, The Netherlands) workstation with ROI measurement tools. Flux was measured for basilar artery and both internal carotid arteries separately. Total flux was calculated for all three vessels together.

pCASL quantitative analysis

MR images were analyzed using FMRIB Software Library (FSL) (Analysis Group, FMRIB, Oxford, UK) [23]. The perfusion was measured in the partial volume corrected gray matter volume. pCASL scans were performed in a total time of 6 minutes. pCASL images were first corrected for slice timing, gradient nonlinearities and subject motion using in-house software and different tools of the FMRIB Software Library (FSL) version 5.0.6.

Files were smoothed and after subtraction a mean CBF over the entire 6- minute pCASL scan was calculated, yielding one average perfusion map for the 6- minute scan. This map was then translated to MNI152 space using FLIRT and FNIRT with the conversion matrix based on the 3DT1 scans. The perfusion was measured in the corrected gray matter volume. This corrected gray matter was segmented in four cortical areas based on the Harvard Oxford probabilistic cortical atlas threshold of 25 percent [24]. These ROI templates were subsequently used with each subject’s gray matter mask to calculate the mean gray matter CBF in units of mL/100g/min for each anatomical region. Perfusion in the hypothalamus and thalamus was calculated in a similar way by measuring perfusion on the CBF maps in thalamic/hypothalamic ROI using Harvard oxford atlas ROI templates for these regions.

One participant was noticed to have extremely tortious blood vessels. For this participant, it was difficult to draw a consistent region of interest for phase contrast MR Angiography. For same reason, he also had strongly decreased ASL labelling efficiency. He was excluded from subsequent analyses.

Statistical Analysis

Descriptive statistics were used to summarize the characteristics of both study groups. Homeostatic model assessment of insulin resistance (HOMA-IR) was calculated by the product of the fasting insulin level (mU/L) and the fasting glucose level (mmol/L) divided by 22.5.

The relation between intranasal insulin and cerebral blood flow was determined using paired t-tests, and presented as means with standard deviation with corresponding p-values. Homogeneity of variance assumption was tested using Levene’s test. Correction for inter- individual and intra- individual differences in perfusion was done by calculating normalized gray matter volumes according to MNI standard space registration. For statistical analyses, Statistical Package for Social Sciences (SPSS) software for windows (version 20.0) was used. Statistical significance was set as P< 0.007, which is the corrected threshold for significance after Bonferroni correction for multiple comparisons.