The effect of intranasal insulin on appetite and mood in obese and non-obese women: an experimental drug study

demographics

Obese women had a higher BMI and self-reported more restrained and emotional eating in line with previous research (eg. [38, 39]† See additional table 2 for participant demographics.

Cookie Intake

There was an important effect of IN insulin on biscuit intake (f(1.50) = 4.59, p= 0.04,p2= 0.08) and follow-up tests showed a reduced biscuit intake for obese women (t(16) = −2.12, p= 0.05, d= 0.46) and no effect for skinny women (t(34) = −0.55, p= 0.59, d.07). There was an important effect of IN insulin on initial ratings of cookie pleasantness ( f(1.47) = 4.83,p= 0.03,p2= 0.09) and follow-up tests showed that IN insulin reduced initial cookie preference in obese women ( t(15) =-2.87, p= 0.01,d= 0.42) but not for skinny women ( t(32) = −0.34, p= 0.74, d= 0.04). There was no main effect of IN insulin on the assessed pleasantness at the end of the session (F(1.47) = 0.01,p= 0.99,p2< 0.01). There was no effect of IN insulin on eating rate ( f(1, 47) = 0.02,p= 0.89,p2< 0.01) (Fig. 2

fig. 2: Eating-related measures.
Figure 2

A represents the biscuit intake of lean and obese women in the placebo (light gray filling) and insulin conditions (dark gray filling). B represents the eating rate of lean and obese women in the placebo and insulin conditions. C shows palatability ratings of lean and obese women at the start of the snack in the placebo and insulin conditions. d shows palatability ratings of lean and obese women at the end of the snack in the placebo and insulin conditions. Squares indicate the mean of each outcome. Asterisks indicate follow-up significance at the 0.05 level.

Appetite and mood

IN insulin decreased appetite ( f(1.50) = 5.66,p= 0.02,p2= 0.10) and this effect was more pronounced for obese women ( t(16) = −2.11, p= 0.051,d= 0.60) than for skinny women ( t(34) = 0.74, p= 0.47,d= 0.11). There were no significant main effects of IN insulin on arousal ( f(1.50) = 2.16,p= 0.15,p2= 0.04), negative effects ( f(1.50) = 1.39,p= 0.24,p2= 0.03), physical effects ( f(1.50) = 0.06,p= 0.81,p2< 0.01) and thirst ( f(1.50) = 0.01,p= 0.98,p2< 0.01).

Also Read:  Can weekly prednisone treat obesity?

There was no significant main effect of IN insulin on PANAS ratings ( f(1.47) = 3.78,p= 0.06,p2= 0.07), but the interaction between IN insulin and BMI was significant ( f(1.47) = 5.47,p= 0.02,p2= 0.10), which was explained by a significant increase in positive affect (PA) ratings for obese women ( t(16) = 2.86, p= 0.01,d= 0.42) (Fig. 3

Fig. 3: Appetite and mood.
figure 3

A represents the post-dose area under the curve (AUC) visual analog scale (VAS) ratings of lean women between placebo (light gray filling) and insulin (dark gray filling). B reflects post-dose AUC VAS ratings of obese women in the placebo and insulin conditions. C reflects post-dose AUC Positive and Negative Affect (PANAS) ratings of lean women between placebo and insulin condition. d reflects post-dose AUC PANAS ratings of obese women in the placebo and insulin conditions. CM centimeters, Min. Minutes, Neg Effects Negative Effects, Physical Effects Physical Effects. Squares represent the mean of each outcome. Asterisks indicate follow-up significance at the 0.05 level.

Cognitive Tasks

The only effect of IN insulin was improved accuracy for the selection of self-referential positive adjectives and rejection of negative adjectives and slower RT on the n-back task for obese women. See below for statistical results and additional table 3 for descriptive statistics.

VPA

The effect of IN insulin condition on recall accuracy was not significant ( f(1.50) = 0.01,p= 0.93,p2< 0.01).

Also Read:  Correspondence on the article "Gestational body weight gain and risk of low birth weight or macrosomia in women of Japan: a countrywide cohort study"

ETB – ECAT

IN insulin improved accuracy for selection of positive adjectives for self-reference and rejection of negative adjectives for self-reference ( f(1.49) = 6.76,p= 0.01,p2= 0.12). Follow-up tests showed no significant differences from IN insulin on ECAT accuracy for lean women ( t(32) = 1.87, p= 0.07,d= 0.22) nor obese women ( t(16) = 1.72 p= 0.10,d= 0.49). The effect of IN insulin on RT ( f(1.48) = 1.64,p= 0.21,p2= 0.03) was not significant.

ETB – EREC

Neither the effect of IN insulin on recall accuracy ( f(1.50) = 0.67,p= 0.42,p2= 0.01) nor errors ( f(1.50) = 0.15,p= 0.70,p2= 0.03) was significant.

ETB – EMEM

Neither the effect of IN insulin on accuracy ( f(1.34) = 2.13,p= 0.15,p2= 0.06), commission errors ( f(1.50) = 1.20,p= 0.28,p2= 0.02) nor RT ( f(1.49) = 0.72,p= 0.40,p2= 0.02) was significant. The interaction between IN insulin and BMI for accuracy was significant ( f(1.34) = 5.65,p= 0.02,p2= 0.14) but follow-up tests showed no significant effects for either BMI groups (ps > 0.05).

N-back

There was no effect of IN insulin on accuracy ( f(1,27) = 1,08,p= 0.31,p2= 0.04), nor RT ( f(1.45) = 1.89,p= 0.18,p2= 0.04). The interaction between IN insulin and BMI was significant for the RT measurement ( f(1.45) = 12.36,p< 0.01,p2= 0.22). Obese women were slower in the insulin condition than in the placebo condition ( t(15) = 3.33, p= 0.01,d= 0.56).

PRT recall

The effect of IN insulin on recall ( f(1.49) = 0.01,p= 0.94,p2< 0.01) was not significant.

Also Read:  Long-term follow-up reduces the risk of type 2 diabetes

Image rating task

Pleasant reviews

There was no significant effect of IN insulin on the pleasantness scores of food stimuli ( f(1.43) = 1.32p= 0.26,p2= 0.3).

fMRI

Statistically significantly greater (FWE-corrected FWE-corrected) blood oxygen level-dependent (BOLD) responses to food compared to non-food images (condition independent) were observed in the left precuneus ( n voxels= 427, xyz†[−6,51,18]Z = 5.46,p< 0.001), left superior frontal gyrus ( n voxels= 611, xyz†[−6,63,−3]† z= 4.78,p< 0.001), left thalamus ( n voxels= 99, xyz =[−3,−15,6]† z= 5.46,p= 0.031), left orbitofrontal cortex ( n voxels = 121, xyz†[−30,12,9]†z= 4.52,=0.016) and left inferior temporal cortex ( n voxels= 99, xyz†[−30,66,45]† z= 4.16, p= 0.046). Allps FWE corrected.

Below a whole-brain FWE-corrected significance threshold, the BOLD response in the left insula [−45,0,6] when viewing pictures of eating versus not eating was significantly greater in the insulin condition compared to placebo in both BMI groups ( n voxels= 107, z= 4.47, p= 0.038) (see Fig. 4† No region showed a decrease in activity after insulin compared to placebo. There was no interaction between food category and treatment.

Fig. 4: fMRI.
figure 4

Top panel: significant cluster (FWE corrected p< 0,05) in de linker insula voor contrast Insuline > Placebo. Threshold figure at initial uncorrected detection thresholdp< 0.001. Bottom panel: left insula cluster parameter estimates [−45,0,6]† Asterisks indicate a significant difference between placebo (light gray filling) and insulin (dark gray filling) for both lean and obese participants at the 0.05 level. Squares represent the mean.

Blood insulin and glucose

Pre-dose blood glucose did not differ according to BMI ( f(1.49) = 0.97,p= 0.33,p2= 0.02) or test day ( f(1.49) = 0.32,p= 0.58,p2= 0.01) (Supplementary Table 4† IN insulin had no effect on blood glucose ( f(1.48) = 0.10,p= 0.75,p2< 0.01) (Supplementary Table 4

Obese women had higher baseline blood insulin levels than lean women ( f(1.44) = 7.03,p= 0.01,p2= 0.14). Obese women had higher blood insulin concentrations after dosing than lean women ( f(1.44) = 13.04,p< 0.01,p2= 0.23). There was a significant increase in blood insulin following IN insulin 5 minutes post-dose ( t(45) = 3.24, p< 0.01,d= 0.36) (see fig. 5

fig. 5: Blood insulin and glucose.
figure 5

Mean (± standard error of the mean) blood insulin concentration of lean women (left) and obese women (right). Fork image indicates lunch timing and cookie image indicates snack timing. Asterisks indicate follow-up significance at the 0.05 level. Women with and without obesity had higher blood insulin concentrations in the insulin condition 5 minutes after the dose.

Leave a Reply

Your email address will not be published.