*3.3. Influence of Diet on Hippocampal Connectivity*

A total of 98 participants were included in the diffusion imaging analyses, after excluding the 6 participants who were missing data for either the mother's education level (*n* = 2) or who failed quality control for the diffusion imaging data (*n* = 4). The amount of fructose children consumed as a percentage of calories was significantly associated with an increase in the mean diffusivity (MD) of the right cingulum, prefrontal connections (ß = 2.51 <sup>×</sup> 10−6, sr = 0.21, *p* < 0.04; Model 1) (Table 4; Figure 1C). Results remained significant after adjusting for age and sex (ß = 2.66 <sup>×</sup> 10−6, sr = 0.22, *<sup>p</sup>* <sup>&</sup>lt; 0.02; Model 2), further adjusting for child BMI z-score (ß = 2.79 <sup>×</sup> 10<sup>−</sup>6, sr = 0.23, *p* < 0.02; Model 3), and additionally adjusting for SES (ß = 3.51 <sup>×</sup> 10−6, sr = 0.28, *p* < 0.003; Model 4), as well as in fully adjusted models (ß <sup>=</sup> 3.64 <sup>×</sup> <sup>10</sup><sup>−</sup>6, sr <sup>=</sup> 0.29, *<sup>p</sup>* <sup>&</sup>lt; 0.002; Model 5). The outcome of the hierarchical modeling indicated that Model 4 explained significantly more variance than Models 1–3, but Model 5 did not explain more variance than Model 4. Therefore, Model 4 was used for the planned post-hoc analyses. The MD for all other tracts and the FA for all the tracts were non-significant. None of the models tested showed significant interactions of sex with dietary intake, nor did they show significant effects of Tanner stage.

Planned post-hoc analyses on the effects of percent of calories from added sugar on the MD of the right cingulum, prefrontal connections resulted in a significant model overall (R<sup>2</sup> = 0.167, F(7,90) = 2.58, *p* < 0.02). Although the overall model was significant, the covariate of interest (i.e., percent calories from added sugar) was not significantly associated with MD of the right cingulum, prefrontal connections (ß <sup>=</sup> <sup>−</sup>2.56 <sup>×</sup> <sup>10</sup><sup>−</sup>7, sr <sup>=</sup> <sup>−</sup>0.07, *<sup>p</sup>* <sup>&</sup>lt; 0.50), indicating that the overall model for added sugar associations with MD of the right cingulum, prefrontal connections was driven by significant associations with decreased age and increased family income. Planned post-hoc analyses on the effects of percent of calories from glucose on the MD of the right cingulum, prefrontal connections also resulted in a significant model overall (R2 = 0.184, F(7,90) = 2.89, *p* < 0.009). However, similar to findings with added sugar, there was not a significant association of percent calories from glucose on the MD of the right cingulum, prefrontal connections (ß = 2.25 <sup>×</sup> 10−6, sr = 0.15, *p* < 0.13). Likewise, similar to

added sugar, the overall model for glucose associations with MD of the right cingulum, prefrontal connections was driven by significant associations with decreased age and increased family income. These result are different from the effects we observed in the volume of the right CA2/3 because MD was only associated with fructose intake and not added sugar or glucose intake.


**Table 4.** Effects of fructose consumption on the mean diffusivity (MD) of right cingulum, prefrontal connections.

Model 1 includes percent calories from fructose only; Model 2 includes age and sex; Model 3 includes BMI z-score; Model 4 includes family income and mom's education; Model 5 includes maternal GDM and maternal pre-pregnancy BMI. Successive models include all covariates from earlier models. BMI: body mass index; LN: <high school; SC: some college; CN: college and post-graduate; GDM: gestational diabetes mellitus; sr = semi-partial r; \* *p* < 0.05; \*\* *p* < 0.01.
