*3.1. Participant Characteristics*

A total of 103 children participated in this study. Child demographic and clinical characteristics are reported in Table 1. The mean ± SD age of the children was 8.55 ± 1.03 years old, 91% of the children were pre-pubertal (Tanner Stage <2), and 62% were girls. Children's BMI ranged from 13.62 to 34.01 kg/m2; BMI percentiles ranged from 5.28 to 99.58; BMI z-scores ranged from <sup>−</sup>1.78 to 2.64. According to CDC standards, 61 (59%) children were healthy-weight, 16 (15%) children were overweight, and 26 (25%) children were obese. Overall, the participants consumed an average (±SD) of 1763 ± 359 kcal per day; 4.57 ± 2.19% of their total calories came from fructose, 4.32 ± 1.81% of their total calories came from glucose, and 13.91 ± 6.89% of total calories from added sugar. Total energy and added sugar intake in our cohort are in line with the general population of the children in the United States [56]. Familial demographics, including maternal education, family income, and mother's self-reported race/ethnicity, are shown in Table 2. Maternal pre-pregnancy BMI ranged from 18.97 to 50.38 kg/m2, and 58% of mothers had GDM during pregnancy.



<sup>1</sup> Percentages were rounded to the nearest percent, therefore sum of variables do not equal 100%; BMI: body mass index.

**Table 2.** Familial demographics.


<sup>1</sup> Percentages were rounded to the nearest percent, therefore sum of variables do not equal 100%; <sup>2</sup> Missing maternal education and family income data from 2 participants.

## *3.2. Influence of Diet on Hippocampal Volumes*

A total of 101 participants were included in the hippocampal volume analyses, after excluding the two participants who were missing data for the mother's education level. The amount of fructose children consumed as a percentage of calories in their diet was significantly associated with an increase in the volume of the right CA2/3 hippocampal subfield (ß = 3.34, sr = 0.25, *p* < 0.01; Model 1)) (Table 3; Figure 1B). Results remained significant after adjusting for ICV, age, and sex (ß = 2.56, sr = 0.19, *p* < 0.02; Model 2), further adjusting for child BMI z-score (ß = 2.80, sr = 0.21, *p* < 0.01; Model 3), and additionally adjusting for SES (ß = 2.70, sr = 0.19, *p* < 0.02; Model 4), as well as in fully adjusted models (ß = 3.33, sr = 0.24, *p* < 0.003; Model 5). These findings suggest that dietary fructose has an effect on hippocampal CA 2/3 volume that is independent of child's age, sex, BMI Z-score, SES, or prenatal exposures. Of note, Model 2, which included covariates for intracranial volume, age, and sex, explained significantly more variance than the unadjusted Model 1, which only contained a covariate for percentage of calories from fructose (F(3,96) = 16.78, *p* < 0.001). This result is unsurprising because the volume of the hippocampus and its subfields are highly dependent upon total brain volume. Model 5, which included all covariates, explained significantly more variance than Model 4, which included all covariates except maternal GDM status and maternal pre-pregnancy BMI (F(2,90) = 6.56, *p* < 0.002). All other hippocampal subfield volumes 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. Additionally, neither fructose (ß = 6893, *p* < 0.30, R<sup>2</sup> = 0.0008), glucose (ß = 7583, *p* < 0.30, R<sup>2</sup> = 0.0008), nor added sugar consumption (ß <sup>=</sup> <sup>−</sup>277, *<sup>p</sup>* <sup>&</sup>lt; 0.89, R2 <sup>=</sup> <sup>−</sup>0.0097) were significantly associated with intracranial volume.


**Table 3.** Associations between dietary fructose intake and the volume of right CA2/3.

Model 1 is percent calories from fructose only; Model 2 includes age, sex and intracranial volume; Model 3 includes BMI z-score; Model 4 includes family income and mom's education; Model 5 includes prenatal exposures to 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; \*\*\* *p* < 0.001.

In planned post-hoc analyses, we explored whether the associations between dietary fructose intake and hippocampal CA 2/3 volume were specific to fructose, or if alternatively the associations were also observed between dietary added sugar and/or glucose intake and CA 2/3 volume. We found that associations between percent of calories from added sugar and volume of the right CA2/3 were also significant (R2 = 0.448, F(10,90) = 7.29, *p* < 0.001). Likewise, planned post-hoc analyses on the effects of percent of calories from glucose on the volume of the right CA2/3 were also significant (R<sup>2</sup> = 0.471, F(10,90) = 8.00, *p* < 0.001). These findings indicated that effects of dietary fructose, added sugar, and glucose intake on right CA2/3 volume were all similar suggesting that increases in the volume of right CA2/3 were non-specific with regards to whether the increased sugar intake was from glucose, fructose, or added sugars. In fact, the effect sizes for glucose intake (sr = 0.27), fructose intake (sr = 0.25), and added sugar intake (sr = 0.23) were nearly identical, indicating each of these types of sugar explained a similar amount of variance. The percent calories from fructose and percent calories from glucose were highly correlated (r = 0.87, *p* < 0.001), and there were moderate correlations between percent calories from fructose and percent calories from added sugar (r = 0.31, *p* < 0.001) and between percent calories

from glucose and percent calories from added sugar (r = 0.48, *p* < 0.001). Given that dietary intake of fructose, glucose, and added sugar were correlated, we did not include them in the same model to test independent effects. It is important to note that added sugars are classified as sugars that are added to foods or beverages when they are processed or prepared, whereas naturally occurring sugars, such as those in fruit or milk, are not classified as added sugars [57].

**Figure 1.** Associations with fructose consumption. (**A**) Using regions of interest for the hippocampus (orange) and its major connections (fornix: pink, uncinate: purple, and the cingulum, which was segmented into the temporal part: yellow and the prefrontal part: turquoise), we identified associations between fructose consumption and the volume of the right CA2/3 subfield of the hippocampus (**B**) and the mean diffusivity of the right cingulum, prefrontal connections (**C**). (**B**,**C**) Graphs show the unadjusted model with no covariates. Solid line indicates the best fit linear trend, with dotted lines showing the confidence interval. Notably, these graphs show the results from Model 1 (Tables 1 and 2, respectively).
