Diet modulates brain network stability, a biomarker for brain aging, in young adults

Abstract

Epidemiological studies suggest that insulin resistance accelerates progression of age-based cognitive impairment, which neuroimaging has linked to brain glucose hypometabolism. As cellular inputs, ketones increase Gibbs free energy change for ATP by 27% compared to glucose. Here we test whether dietary changes are capable of modulating sustained functional communication between brain regions (network stability) by changing their predominant dietary fuel from glucose to ketones. We first established network stability as a biomarker for brain aging using two large-scale (n = 292, ages 20 to 85 y; n = 636, ages 18 to 88 y) 3 T functional MRI (fMRI) datasets. To determine whether diet can influence brain network stability, we additionally scanned 42 adults, age < 50 y, using ultrahigh-field (7 T) ultrafast (802 ms) fMRI optimized for single-participant-level detection sensitivity. One cohort was scanned under standard diet, overnight fasting, and ketogenic diet conditions. To isolate the impact of fuel type, an independent overnight fasted cohort was scanned before and after administration of a calorie-matched glucose and exogenous ketone ester (d-β-hydroxybutyrate) bolus. Across the life span, brain network destabilization correlated with decreased brain activity and cognitive acuity. Effects emerged at 47 y, with the most rapid degeneration occurring at 60 y. Networks were destabilized by glucose and stabilized by ketones, irrespective of whether ketosis was achieved with a ketogenic diet or exogenous ketone ester. Together, our results suggest that brain network destabilization may reflect early signs of hypometabolism, associated with dementia. Dietary interventions resulting in ketone utilization increase available energy and thus may show potential in protecting the aging brain.

Keywords: beta-hydroxybutyrate; brain; glucose; ketone; neural.

Fig. 1.

Brain networks destabilize with age, with the strongest impact in the auditory, higher visual processing (V2), and basal ganglia networks (total n = 928). (A) Leipzig Mind-Brain-Body open-source dataset, ages 20 to 85, binarized into younger (n = 214) vs. older (n = 78) participants (Mann–Whitney U = 0.28; P = 1.4 × 10⁻⁸) and Cam-CAN open-source dataset, ages 18 to 88, binarized into younger (n = 281) vs. older (n = 355) participants (Mann–Whitney U test = 0.27, P = 1.6 × 10⁻²²). We fit network instability for the Cam-CAN dataset using a (logistic) sigmoidal function (nonlinear least squares with weights inversely proportional to the SD, reduced χ² = 1.07). From this fit, we obtained the inflection point (switch point), which occurs at 60 ± 2 y, and the width accounting for 90% of the transition of 13 ± 6 y, resulting in an onset of degeneration at 47 y. A linear fit to the data resulted in a 30% larger reduced χ² value, indicating that the data are more accurately fitted by a sigmoidal rather than linear fit. (B) Increasing brain age, defined by network stability, predicts progressively lower cognition (MMSE scores). Linear fit to brain age vs. MMSE score data finds a slope of –0.66 ± 0.27 (estimate ± SE), implying instability-derived brain age increases 0.66 y for every point decrease in MMSE score (P < 0.01, CI = [–1.18, –0.14]). For younger individuals, T2D accelerates brain aging compared to age-matched healthy controls. Mean actual ages for younger individuals with (51 ± 4 y, n = 14) and without (51 ± 5 y, n = 109) T2D were equivalent, while brain age for young diabetics was significantly increased over that of healthy controls (younger T2D vs. HC, Mann–Whitney U = 0.34, P = 0.0002). For older individuals, mean actual ages for T2D (73 ± 6 y, n = 14) and for HC (74 ± 6 y, n = 82) were equivalent to their respective brain ages (older T2D vs. HC, Mann–Whitney U test = 0.48, P = 0.87).

Fig. 2.

Brain networks destabilize with glucose and stabilize with ketones. (A) In the diet experiment, each participant was scanned three separate times, time locked to eliminate diurnal variability: while following a standard diet (STD), after overnight fasting, and after following a ketogenic diet for 1 wk (τ = 1, repeated-measures ANOVA LSD post hoc, standard vs. ketogenic diet: t = 5.4 P = 0.0000001). (B) To isolate fuel source as the variable of interest between the diets, we followed up with a bolus experiment. Each participant was scanned two separate times, again time locked to eliminate diurnal variability, with the d-βHb ketone ester individually weight dosed (395 mg/kg). Each individual’s glucose dose was then calorie matched to his or her d-βHb ketone ester dose. For each session we subtracted intrasession fasting values from each bolus value (τ = 1, paired t test, glucose bolus minus fasting vs. ketone ester bolus minus fasting: t = 2.9, P = 0.004). (C) The ketone ester’s stabilizing effects were observed even under high glycemic load; here we show network stability values for a single participant, following a standard diet that included a 75 g glucose challenge, with and without administration of the ketone ester (τ = 1, paired t test, high-glycemic standard diet with vs. without 25 g d-βHb ketone ester bolus: t = 4.12, P = 0.0001). Error bars for the case study (n = 1) reflect statistics calculated over up to 24 windows for τ = 1, 23 windows for τ = 2, etc. Equivalent effects for the same participant performing motor and spatial navigation tasks are shown in SI Appendix, Fig. S4. n.s., not statistically significant; *P ≤ 0.05; **P ≤ 0.01; ***P ≤ 0.0001.

Fig. 3.

ALFF, a general measure of brain activity, was increased for participants on the ketogenic diet compared to their standard (std) diets (n = 12). This remained true for resting state, as well as during motor and spatial navigation tasks. Resting state and motor tasks were of 10 min duration. Spatial navigation shows the first 10 min (for comparison with other tasks) and then an additional 30 min, for 40 min total. This was done to assess fatigue effects over longer periods of time. Comparing symmetry over time between shifts from lower- to higher-activity states versus shifts from higher- to lower-activity states, both the ketogenic and fasting conditions showed a mean of zero bias (one-sample t test ketogenic diet: t = –0.22, P = 0.83; overnight fast: t = 0.26, P = 0.80), whereas the standard diet condition showed the brain switching from high- to lower-activity states (standard diet: t = –3.29, P = 0.007).