ArticlePDF Available

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



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.
Diet modulates brain network stability, a biomarker for
brain aging, in young adults
Lilianne R. Mujica-Parodi
, Anar Amgalan
, Syed Fahad Sultan
, Botond Antal
, Xiaofei Sun
, Steven Skiena
Andrew Lithen
, Noor Adra
, Eva-Maria Ratai
, Corey Weistuch
, Sindhuja Tirumalai Govindarajan
Helmut H. Strey
, Ken A. Dill
, Steven M. Stufflebeam
, Richard L. Veech
, and Kieran Clarke
Department of Biomedical Engineering, Stony Brook University, Stony Brook, NY 11794;
Laufer Center for Physical and Quantitative Biology, Stony Brook
University, Stony Brook, NY 11794;
Department of Physics and Astronomy, Stony Brook University, Stony Brook, NY 11794;
Athinoula A. Martinos Center
for Biomedical Imaging, Massachusetts General Hospital and Harvard Medical School, Charlestown, MA 02129;
Department of Computer Science, Stony
Brook University, Stony Brook, NY 11794;
Department of Applied Mathematics and Statistics, Stony Brook University, Stony Brook, NY 11794;
of Metabolic Control, NIH/National Institute on Alcohol Abuse and Alcoholism, Rockville, MD 20852; and
Department of Physiology, Anatomy, and
Genetics, University of Oxford, Oxford OX1 3PT, United Kingdom
Contributed by Ken A. Dill, January 9, 2020 (sent for review July 30, 2019; reviewed by Peter Crawford and Stephen C. Cunnane)
Epidemiological studies suggest that insulin resistance accelerates
progression of age-based cognitive impairment, which neuroimag-
ing has linked to brain glucose hypometabolism. As cellular inputs,
ketones increase Gibbs free energy change for ATP by 27% com-
pared to glucose. Here we test whether dietary changes are capa-
ble 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 administra-
tion 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 cogni-
tive acuity. Effects emerged at 47 y, with the most rapid degener-
ation 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. To-
gether, our results suggest that brain network destabilization
may reflect early signs of hypometabolism, associated with demen-
tia. Dietary interventions resulting in ketone utilization increase
available energy and thus may show potential in protecting
the aging brain.
Because the human brain is only 2% of the bodys volume but
consumes over 20% of its energy (1, 2), it is particularly
vulnerable to changes in metabolism. Dietary increase in glyce-
mic load over the past 100 y has led to a national epidemic of
insulin resistance (type 2 diabetes [T2D]) (3, 4), which has been
identified by several large-scale epidemiological studies as an
early risk factor for later-life dementia (5). For example, a post
hoc analysis of the UK Whitehall II cohort study (n=5,653)
reported that those with diabetes showed a 45% faster decline in
memory, a 29% faster decline in reasoning, and a 24% faster
decline in global cognitive score and that the risk of accelerated
cognitive decline in middle-aged patients with T2D is dependent
on both disease duration and glycemic control (6). Similar results
were reported using cohorts obtained from Israel (n=897) (7)
and the United States (n=4,135) (8), the latter of which found
the relationship between T2D and cognitive dysfunction to be
evident even in younger adults. This marked association has led
some researchers to propose that dementia may be the brains
manifestation of metabolic disease (9).
This association is all the more surprising because, until quite
recently, the brain was assumed to make use of purely insulin-
independent transport of glucose into cells (GLUT3), utilizing
neither insulin nor insulin transport (GLUT4). However, there
now is rapidly accumulating evidence that insulin is directly
relevant to neurons, brain aging, and associated memory deficits.
For example, an early breakthrough study with radioactive insulin
staining found that, contrary to the assumption that neurons did
To better understand how diet influences brain aging, we focus
here on the presymptomatic period during which prevention
may be most effective. Large-scale life span neuroimaging
datasets show functional communication between brain regions
destabilizes with age, typically starting in the late 40s, and that
destabilization correlates with poorer cognition and accelerates
with insulin resistance. Targeted experiments show that this
biomarker for brain aging is reliably modulated with consump-
tion of different fuel sources: Glucose decreases, and ketones
increase the stability of brain networks. This effect replicated
across both changes to total diet as well as fuel-specific calorie-
matched bolus, producing changes in overall brain activity that
suggest that network switchingmayreflectthebrains adaptive
response to conserve energy under resource constraint.
Author contributions: L.R.M.-P., S.S., E.-M.R., K.A.D., R.L.V., and K.C. designed research;
A.A., A.L., N.A., S.T.G., and S.M.S. performed research; L.R.M.-P., A.A., S.F.S., X.S., S.S.,
E.-M.R., C.W., H.H.S., R.L.V., and K.C. contributed new reagents/analytic tools; L.R.M.-P.,
A.A., S.F.S., B.A., X.S., A.L., N.A., E.-M.R., and H.H.S. analyzed data; and L.R.M.-P., A.A.,
S.F.S., B.A., A.L., N.A., E.-M.R., H.H.S., R.L.V., and K.C. wrote the paper.
Reviewers: P.C., University of Minnesota; and S.C.C., Université de Sherbrooke.
Competing interest sta tement: The intellect ual property covering the uses of keto ne
bodies and ketone esters is owned by BTG Plc., Oxford University Innovation Ltd., and
the NIH. R.L.V. and K.C., as inventors, will receive a share of the royalties under the terms
prescribed by each institution. K.C. is a director of TΔS Ltd., a company spun out of the
University of Oxford to develop products based on the science of ketone bodies in human
nutrition. TΔS Ltd. has licensed HVMN Inc. to sell the ketone ester in sports drinks in the
United States.
This open access article is distributed under Creative Commons Attribution License 4.0
(CC BY).
Data deposition: All datasets are located at Data Archive for the Brain Initiative (DABI;
https://dabi.l ore/project/42 ) in the Protecti ng the Aging Brain (P AgB),
Project 1926781 repository. Additional details (including links to code used in the process-
ing and analyses of data) can be found at
L.R.M.-P. and A.A. contributed equally to this work.
To whom correspondence may be addressed. Email: or
Deceased February 2, 2020.
This article contains supporting information online at
doi:10.1073/pnas.1913042117/-/DCSupplemental. PNAS Latest Articles
Downloaded by guest on March 3, 2020
not utilize insulin, the rat brain was, in fact, densely populated
with insulin receptors in both the hippocampus and cortex (10).
Positron emission tomography in humans has demonstrated re-
duced glucose uptake in insulin-resistant participants versus healthy
controls (11), suggesting that even the earliest stages of T2D in-
duce hypometabolism of neurons, as with other cells in the body
and as per brain glucose hypometabolism commonly seen in de-
mentia. Finally, infusing insulin, without increasing glucose, has
been shown to increase memory for Alzheimers disease patients
(12). These clinical studies suggest that deleterious cognitive ef-
fects of insulin resistance may result from metabolic stress, as
neurons gradually lose access to glucose. If so, it may be possible
to bypass insulin resistance to refeed neurons by exploiting ketone
bodies as an alternative fuel.
Endogenous ketone bodies, including D-β-hydroxybutyrate, are
primarily produced in the liver from long- and medium-chain free
fatty acids released from adipose tissue during hypocaloric/fasting
states or food when following a low-carbohydrate/moderate-
protein/high-fat diet (13). In rats, neurological and cognitive
effects of glucocorticoid-induced insulin resistance in the hip-
pocampus were reversed by ketone bodies (D-β-hydroxybutyrate)
and mannose but not by either glucose or fructose (14). Like-
wise, in humans there is evidence that even as older brains
become hypometabolic to glucose, neural uptake of ketone bodies
remains unaffected, even for the most severe glucose hypo-
metabolism endemic to Alzheimers disease (15, 16). Finally,
lifelong hypocalorically induced ketosis preserves synaptic plas-
ticity (17) and cognition (18) in elderly animals [chronological age
equivalent to 87 to 93 human years (19)].
Beyond the ability to short-circuit insulin resistance, however,
ketone bodies have other metabolic advantages (2024) that may
confer neurobiological benefits even to younger healthy indi-
viduals not yet in a deficit (hypometabolic) state. Of those ad-
vantages, one of the most fundamental is that, as cellular inputs,
β-hydroxybutyrate molecules increase Gibbs free energy change
for ATP by 27% compared to glucose (24). While it is currently
unknown how increasing available energy might impact a healthy
brain, one consequence suggested by prior animal data is an in-
crease in neurotransmitter production. Eight- to ten-month-old
mice, the chronological equivalent of 27- to 33-y-old humans
(19), showed increased synaptic efficiency, low-theta band oscilla-
tions, and learning consolidation during intermittent-fasting-induced
ketosis (25). Mechanistically, this increase in synaptic efficiency was
linked to increased expression of the N-methyl-D-aspartate
(NMDA) receptor for glutamate.
Here we test two hypotheses. First, we investigate the time
course of brain aging in humans to determine whether there is
evidence for a long-term degenerative process that lays the foun-
dation for neurometabolic stressdecades before cognitive effects
become evident. This is clinically critical because it identifies a
window of time during which neurodegenerative effects may still be
reversible if we can increase neuronsaccess to fuel. Second, to
isolate the role of energy in modulating this variable, we hold age
constant while testing the neurobiological impact of switching the
primary fuel source of the human brain from glucose to ketone
bodies. The above translational results showed that fasting in-
creases NMDA-driven synaptic efficiency (25); neurotransmission,
in turn, has been shown to drive change in cerebral blood flow (26)
and thus functional communication between brain regions mea-
sured by blood oxygen leveldependent (BOLD) functional MRI
(fMRI) resting-state connectivity (27). Therefore, we expected that
ketone bodies might improve fMRI-derived measures of neurobi-
ological functioning, even in healthy younger adults.
To test these hypotheses, we proceeded in two stages. First,
using independent large-scale human fMRI datasets, sampling
across the adult life span (ages 18 to 88), we established a whole-
brain-scale biomarker (network stability, defined as the brains
ability to sustain functional communication between its regions)
that robustly associates with brain aging. Second, we conducted
two targeted experiments in humans, optimized for detection
sensitivity at the single-participant level, to test the impact of
manipulating fuel type: glucose versus ketone bodies, using both
diet and bolus, on that biomarker. Of note, while translational
studies tend to employ long-term (lifelong) dietary modifications
equivalent to 20 to 30 y of human life spanfor our targeted ex-
periments we deliberately focused on rapid effects (after 1 wk
of the ketogenic diet and half an hour for the D-βHb ketone
ester). This was done for three reasons. First, it permitted a
within-subject design, thereby rigorously controlling for genetic
and environmental differences between subjects. Second, it
narrowed down the number of potential biological mechanisms to
those capable of acting over minutes or days, rather than months,
years, or decades. Finally, we maximized clinical relevance by
using dietary modifications that would be realistic to implement by
most individuals in real-world environments.
Life Span Neuroimaging Datasets. To identify network stability across the life
span, we analyzed two large-scale open-source 3 T fMRI resting-state
datasets: Max Planck Institute Leipzig Mind-Brain-Body (28) (Leipzig: ages
20 to 85, n=292) and Cambridge Centre for Ageing and Neuroscience Stage
II (29) (Cam-CAN: ages 18 to 88, n=636). Leipzig showed a bimodal distri-
bution of individuals older and younger than 50, which required statistical
analyses of age as a discrete variable. Cam-CAN sampled more evenly across
the life span, permitting additional statistical analyses of age as a continuous
Metabolic Neuroimaging Datasets. To determine whether fuel affects brain
network stability, we conducted resting-state scans on two independent
cohorts of young healthy adults. Subjects were asked to keep their eyes open
and let their minds wander while focusing on a white orienting cross on an
otherwise black screen. To achieve the higher signal/noise required to ana-
lyze data at the single-participant level, participants were scanned using
ultrahigh-field (7 T) fMRI at the Massachusetts General Hospital Athinoula A.
Martinos Center for Biomedical Imaging, using acquisition parameters
quantitatively optimized via dynamic phantom for detection sensitivity to
resting-state networks (30). Immediately prior to and following each scan,
blood glucose and ketone (D-β-hydroxybutyrate) levels were measured using
Precision Xtra test strips (Abbott Laboratories) (Table 1). Exclusion criteria for
all three studies included MR contraindications for ultrahigh-field imaging;
diagnoses of psychiatric and/or neurological disorders; traumatic brain in-
juries; recreational drug usage, including severe alcohol use; and/or (for
females) pregnancy. Participants were excluded if they were currently fol-
lowing or had recently followed (within past 6 mo) a low-carbohydrate or
ketogenic diet. Detailed clinical and demographic characteristics for all in-
dividuals participating in the metabolic studies can be found in SI Appendix,
Table S1. Studies were registered as a clinical trial on
(identifier NCT04106882) and approved by the institutional review boards of
Massachusetts General Hospital (Partners Healthcare) and Stony Brook
University; all participants provided informed consent. For access to relevant
datasets (31) and code used to process and analyze the data, see Datasets
For the first experiment (diet) (n=12, μ
=28 ±7 y; 4 female), we
scanned participants under three conditions: 1) standard diet: following
their standard diet, without fasting; 2) fasting: following their standard diet,
with an overnight (12 h) fast; and 3) ketogenic diet: following a ketogenic
(high-fat, moderate-protein, low-carbohydrate [<50 g/d]) diet for 1 wk, by
which point all participants were in ketosis (>0.6 mmol/L ketone blood
For the second experiment (bolus) (n=30, μ
=29 ±8 y; 18 female), we
scanned an independent cohort of participants under three conditions: 1)
fasting: following their standard diet, with an overnight fast; 2) glucose
bolus: breaking the fast with a glucose drink (Glucose Tolerance Test Bev-
erages, Fisher Scientific Inc.); and 3) D-βHb ketone ester bolus: breaking the
fast with a ketone drink (D-β-hydroxybutyrate ketone ester; HVMN).
The D-βHb ketone ester was weight dosed for each participant at 395
mg/kg and calorically matched (μ
=125 ±19) between D-βHb ketone ester
=26.65 ±3.97 g) and glucose (μ
=31.33 ±4.57 g). Prior to
neuroimaging, we acquired fasting plasma glucose and insulin measures
for calculation of insulin resistance using HbA1c (μ
=5.14 ±0.32%
[min/max =4.6 to 5.8%; insulin resistant >5.6%]) and the Homeostatic Model
| Mujica-Parodi et al.
Downloaded by guest on March 3, 2020
Assessment of Insulin Resistance (HOMA-IR) (μ
=1.41 ±0.59 [min/max =
0.41 to 2.87; insulin resistant >2.0]). HOMA-IR was calculated as fasting insulin
(μU/mL) ×fasting glucose (mg/dL)/405 (32).
While ketone pharmacokinetics for human peripheral blood concentra-
tions have been established (33), ketone pharmacokinetics for the human
brain were unknown. Thus, to establish optimal timing for the bolus study,
we first performed a magnetic resonance spectroscopy (MRS) study (n=8;
=27 ±5 y; 3 female) to determine the bolus time course in the brain.
Using a within-participant time-locked design, as well as weight- and calorie-
matched dosing as described below, we measured brain glucose and
β-hydroxybutyrate at baseline, then every 5 min for 90 min after adminis-
tering each bolus. As per SI Appendix, Fig. S2,MRS showed glucose and
ketones reaching peak concentrations in the brain at 30 min postbolus. Of
the two fuel types, glucose was confirmed to be shorter acting and more
volatile compared to ketones (postpeak coefficient of variation was 2.1 ±0.8
for glucose and 0.14 ±0.03 for the D-βHb ketone ester; P=0.04), which
remained at their peak for at least 90 min postbolus. Thus, to ensure peak
concentrations in the brain for both glucose and ketones, for the bolus ex-
periment we acquired 10 min resting-state scans starting 30 min postbolus.
To check for potential interactions between diet and bolus conditions, as
well as to test whether the D-βHb ketone ester could, in principle, counteract
the effects of higher glycemic load, we conducted an additional investiga-
tion using one participant (case study: female, age 47, HbA1c =5.8%). For
the case study, the baseline condition consisted of a standard diet supple-
mented 30 min prior to the scan with a 75 g glucose bolus, a standardized
challenge dose used clinically for the oral glucose tolerance test (34). In a
time-locked within-subject design, the participant was scanned twice: on
one day with a weight-dosed (395 mg/kg) 25 g D-βHb ketone ester bolus and
on another day without it. Each of these two conditions was conducted at
resting state and while performing spatial navigation and motor tasks, as
described below.
MRI Acquisition. Both life span datasets were acquired at 3 T field strength;
Leipzig had a time to repetition (TR) =1,400 ms over 15 min and 30 s, while
Cam-CAN had a TR =1,970 ms over 8 min and 40 s [further details may be
found in dataset documentation (28, 29)]. Given the focus on clinical ap-
plications, requiring single-participant-level resolution, all metabolic
datasets were acquired at ultrahigh-field (7 T) field strength and included
whole-brain BOLD (echoplanar imaging, EPI), field map, and T1-weighted
structural (multi-echo magnetization prepared rapid gradient echo
[MEMPRAGE]) images. BOLD images were acquired using a protocol quan-
titatively optimized, using a dynamic phantom (BrainDancer; ALA Scientific
Instruments), for detection sensitivity to resting-state networks (30): Simul-
taneous multi-slice (SMS) slice acceleration factor =5, R=2 acceleration in
the primary phase encoding direction (62 reference lines) and online gen-
eralized autocalibrating partially parallel acquisition (GRAPPA) image re-
construction, TR =802 ms, echo time (TE) =20 ms, flip angle =33°, voxel
size =2×2×1.5 mm, slices =85, and number of measurements =2,320 in
each of the prebolus and postbolus intervals, for a total acquisition time of
62 min. Field map images were acquired using the following parameters:
TR =723 ms, TE1 =4.60 ms, TE2 =5.62 ms, flip angle =36°, voxel size =1.7 ×
1.7 ×1.5 mm, and slices =89, for a total acquisition time of 3 min, 14 s. The
whole-brain T1-weighted structural volumes were acquired with 1 mm iso-
tropic voxel size and four echoes with the following protocol parameters:
TE1 =1.61 ms, TE2 =3.47 ms, TE3 =5.33 ms, TE4 =7.19 ms, TR =2,530 ms,
and flip angle =7°, with R=2 acceleration in the primary phase encoding
direction (32 reference lines) and online GRAPPA image reconstruction, for a
total volume acquisition time of 6 min, 3 s.
Spatial Navigation and Motor Tasks. To assess whether effects extended be-
yond resting state to tasks that increased cognitive load and therefore brain
metabolic demand, for the diet study and case study, participants additionally
navigated virtual reality mazes using an MR-compatible joystick (Nata
Technologies). We created these mazes using the AldousBroder algorithm in
Daedalus ( and programmed
them for a virtual reality scanner environment using Vizard (WorldViz). For
the spatial navigation task, participants made use of spatial encoding and
memory in finding their way from one end of the maze and back. For the
motor task, participants simply followed a corridor and therefore navi-
gated without making decisions.
MRI Preprocessing. Life span preprocessing was conducted in the FMRIB
Software Library (FSL; Anatomical im-
ages were skull stripped and coregistered to Montreal Neurological Institute
(MNI) templates and mean functional images. Functional images were mo-
tion and field map corrected, brain extracted, and coregistered to MNI
templates using transformations learned through the anatomical image.
Metabolic preprocessing used Statistical Parametric Mapping 12 (SPM12; combined with an image
processing workflow established with fMRIPrep (35). Anatomical images
(MEMPRAGE) were normalized to MNI templates using unified segmentation
and registration. Images of each individual participant were realigned to ac-
count for head movements and field map corrected for geometric distortions
caused by the magnetic field inhomogeneity, followed by normalization to
MNI space. Physiological confounds were removed using the Component
Based Noise Correction Method (CompCor) (36). No spatial smoothing was
applied to any of the datasets. For all datasets, voxelwise data were parceled
into the Willard 499 functional region of interest (ROI) atlas, which further
coarse grained data into 14 resting-state networks (37).
fMRI Network Analyses. To probe temporal dynamics and reorganization of
communication across brain regions (interregional communication, typically
described as brain networks, and those networkspersistence over time,
defined as network stability), ROI-level fMRI time series were binned into
nonoverlapping time windows, or snapshots,of 24 s. From each window
an all-to-all, signed, symmetric network of correlation strengths was extracted.
We quantified the stability of brain networks in two complementary ways. To
measure gross difference, we calculated total instability, defined as the (scalar)
norm of difference in the correlation matrix for each pair of distinct snapshots
of the brain network, where τis the time duration (in units of 24 s) over which
persistence was calculated (SI Appendix,Fig.S5). To identify which networks
across the brain were most responsible for these effects, we calculated the
least absolute shrinkage and selection operator (LASSO) regression on the fea-
ture set of instabilities calculated from resting-state networks and structural
Table 1. Blood glucose and ketone measurements for MRS time course (n=8) and fMRI bolus (n=30) studies
MRS time course study
Glucose bolus Ketone ester bolus
PRE POST 10 min POST 80 min PRE POST 10 min POST 80 min
Blood glucose,
mg/dL (mmol/L)
91 ±13 (5.1 ±0.7) 95 ±14 (5.3 ±0.8) 89 ±12 (4.9 ±0.7) 89 ±6 (4.9 ±0.3) 82 ±9 (4.6 ±0.5) 67 ±7(3.7±0.4)
Blood βHb, mmol/L 0.2 ±0.1 0.2 ±0.2 0.2 ±0.1 0.1 ±0.1 0.4 ±0.4 3.6 ±0.7
Bolus study
Glucose bolus Ketone ester bolus
PRE POST 10 min POST 50 min PRE POST 10 min POST 50 min
Blood glucose,
mg/dL (mmol/L)
95 ±12 (5.3 ±0.7) 101 ±16 (5.6 ±0.9) 90 ±13 (5.0 ±0.7) 92 ±11 (5.1 ±0.6) 91 ±12 (5.1 ±0.7) 74 ±12 (4.1 ±0.7)
Blood βHb, mmol/L 0.2 ±0.1 0.2 ±0.1 0.1 ±0.1 0.1 ±0.1 1.4 ±1.2 3.5 ±1.1
The Abbott Precision Xtra Glucose & Ketone Monitoring System was used for all fingerstick blood measurements. PRE =prebolus; POST =postbolus.
Mujica-Parodi et al. PNAS Latest Articles
Downloaded by guest on March 3, 2020
parcellations as defined by the Automated Anatomical Labeling atlas (AAL; These identified a data-driven con-
struct, brain age, with the minimum number of coefficients. To measure large-
scale functional reorganization, we calculated module instability for fMRI data
acquired from the diet study (n=12), indicating the extent to which nodes in a
network module switched modules over time. Modules are defined as the
nonoverlapping partition of all nodes in the network, such that intramodule
connections are maximized relative to the intermodule connections. For each
network matrix, modules were extracted using the Louvain parameter-free
modularity-maximization algorithm (38). To obtain the (scalar) module in-
stability, we averaged over all nodes: calculating the percentage of the nodes
neighbors within the same module that failed to remain in the same module for
the next network snapshot. Finally, to quantify general brain activity (39, 40), we
calculated the whole-brain signal amplitude for low-frequency (0.01 to 0.1 Hz)
fluctuations (ALFF).
Brain Networks Destabilize with Age. Across the life span measured
by Cam-CAN, cognitive acuity declined with age, as measured by
the standard clinical instrument used to assess dementia, the
Mini-Mental State Examination (MMSE) (41) (r=0.30, P=
3.63 ×10
). Cam-CAN and Leipzig resting-state datasets show
that increased age, in turn, was associated with destabilization of
brain networks (Leipzig <50 y [n=214] vs. 50 y [n=78], Mann
Whitney Utest =0.28, P=1.4 ×10
;Cam-CAN<50 y [n=281]
vs. 50 y [n=355], MannWhitney Utest =0.27, P=1.6 ×10
Fig. 1 A,Left). This effect was driven primarily by the dynamics of
three resting-state functional networks (37): auditory (superior
temporal gyrus), higher visual processing (V2) and basal ganglia
(thalamus, caudate, inferior frontal gyrus). LASSO regression with
instability of the 12 resting-state networks as predictor variables and
age as the predicted variable identified these three networks with
high selectivity (r=0.30, P=7.11 ×10
), assigning all other net-
works zero weight.
Age-associated degradation in network stability was sigmoidal
(Fig. 1 A,Right;n=636; sigmoid reduced χ
=1.07 vs. linear
reduced χ
=1.39), with an inflection point of 60 ±2 y, indicating
the age at which network stability degraded most precipitously.
The base of the sigmoid was 13 y earlier; thus, networks in our
life span dataset started to destabilize at 47 y. Importantly, this
suggests that the first latent markers for brain aging may be ca-
pable of neurobiological detection decades before cognitive
symptoms become evident.
The three most dominantly affected networks were combined
into a single variable, brain age, linearly composed of Cam-CAN-
derived network stability values for the auditory (β
: 0.25/
1.77), higher visual processing (β
: 0.35/2.51), and basal
ganglia (β
: 0.16/1.19) networks. Brain age inversely cor-
related with cognitive acuity (Fig. 1 B,Left). Moreover, for
younger individuals, T2D accelerated brain aging compared to
age-matched healthy controls (Fig. 1 B,Right). Mean actual ages
for younger individuals with (51 ±4 y) and without (51 ±5y)
T2D were equivalent, while brain age for young diabetics was
significantly higher than that of healthy controls (younger T2D
vs. HC, MannWhitney Utest =0.34, P=0.0002). For older
individuals, both mean actual ages for individuals with (73 ±6y)
and without (74 ±6 y) T2D and brain ages for the two groups
(older T2D vs. HC, MannWhitney Utest =0.48, P=0.87) were
equivalent. Thus, younger individuals with T2D showed brain
network destabilization (i.e., brain age) that, for nondiabetics,
normally would be seen at an older age.
Ketosis Stabilizes Brain Networks. Experimental modulation of
fuel intake shows that brain networks are stabilized in healthy
younger adults through ketosis both induced by a 1 wk change of
diet (τ=1, repeated-measures ANOVA least significant differ-
ence (LSD) post hoc, standard vs. ketogenic diet: t=5.4, P=
0.0000001, n=12; Fig. 2A) and as rapidly as 30 min following
ingestion of exogenous D-βHb ketone ester (τ=1, paired ttest,
glucose bolusfasting vs. D-βHb ketone ester bolusfasting, t=
2.9, P=0.004, n=30; Fig. 2B). Overall, both ketosis induced by
a ketogenic diet and ketosis induced by drinking exogenous D-
βHb ketone ester showed effects equivalent to those seen with
Age-Matched T2D/HC, from Cam-CAN (N=219)
Age-Impacted Networks: Auditory, Higher Visual Processing (V2), Basal Ganglia
Average Instability per Connection Over Shortest ( =1, 24s) Time Delay
Brain Network Instability Over Lifespan
Associated with
Cam-CAN (N=636)
Cam - CAN
(N=636 )
Brain Network Instability (r)
“Brain” vs. Actual Age (Years)
“Brain Age” (Years)
“Brain Age” as Dened by Age-Impacted Networks
Accelerated with
Type 2 Diabetes
Actual Age
“Brain Age”
Younger Older
at Approximately 47 Yrs. of Age
25 26 27 28 29 30
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 (MannWhitney U=0.28;
P=1.4 ×10
) and Cam-CAN open-source dataset, ages 18 to 88, binarized into younger (n=281) vs. older (n=355) participants (MannWhitney Utest =
0.27, P=1.6 ×10
). 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 χ
=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 χ
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 ±4y,n=14) and without (51 ±5y,n=109) T2D were equivalent, while brain
age for young diabetics was significantly increased over that of healthy controls (younger T2D vs. HC, MannWhitney U=0.34, P=0.0002). For older in-
dividuals, mean actual ages for T2D (73 ±6y,n=14) and for HC (74 ±6y,n=82) were equivalent to their respective brain ages (older T2D vs. HC, Mann
Whitney Utest =0.48, P=0.87).
| Mujica-Parodi et al.
Downloaded by guest on March 3, 2020
Change in Brain Network Stability After One Week Ketogenic Diet (N=12)
overnight fast
standard diet
Change in Brain Network Stability 30 Minutes After Drinking Calorie-Matched Glucose vs Ketone Ester (
Bolus (N=30)
Ketone Ester Stabilizes Brain Networks Even After High-Glycemic Load: Standard Diet with High (75g) Glucose (Case Study, N=1)
All Brain Networks
Average Instability per Connection
Over Shortest ( W=1; 24s)
Time Delay
Over All ( W=1-20; 24s-8 min)
Time Delays
overnight fast
overnight fast
30 minutes
Ketone ester (D-EHb) individually weight-dosed at 395 mg/kg.
Glucose dose calorie matched to D-EHb dose.
standard diet
standard diet GLU
30 minutes
D-EHb –
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 individuals 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 ttest, glucose bolus minus fasting vs. ketone ester bolus minus fasting: t=2.9, P=0.004). (C) The ketone esters 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 ttest, 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;**P0.01;***P0.0001.
Mujica-Parodi et al. PNAS Latest Articles
Downloaded by guest on March 3, 2020
fasting (τ=1, repeated-measures ANOVA LSD post hoc, Die-
:P=0.75, n=12; Bolus
:P=0.1, n=30; Fig. 2
Aand B), while a standard diet and glucose bolus consistently
destabilized brain networks. As a measure, network stability
showed robust testretest reliability, with minimal intrasubject
variation across the bolus studys two fasting sessions spaced an
average of 4 d (±2 d) apart (τ=1, repeated-measures ANOVA
LSD post hoc, fasting session 1 vs. fasting session 2, P=0.28,
n=30; SI Appendix,Fig.S3).
Clearly visible even at the single-participant level, our case
study showed the D-βHb ketone ester has brain network stabi-
lizing effects even under high glycemic load (τ=1, paired ttest,
standard diet+75g glucose bolus with vs. without D-βHb ketone
ester bolus, t=4.12, P=0.0001; Fig. 2Cand SI Appendix, Fig.
S4). Blood values for the case study are provided in SI Appendix,
Fig. S2 and Table S2.
Further analyses showed network instability occurs from large-
scale reorganization of network modules (network switching; SI
Appendix,Figs.S5Aand S6 Aand B), rather than changing of
connection strengths while preserving modules (network dimming
or flickering; SI Appendix,Figs.S5Band S6 Cand D). The
fMRI signals ALFF, a general measure of brain activity, was
consistently higheracross both rest and task conditionsfor the
participants following a ketogenic diet or fasting compared to
following their standard diets (resting state: P=1.1 ×10
task: P=1.3 ×10
; spatial navigation [early: 0 to 10 min]: P=
6.7 ×10
; spatial navigation [late: 10 to 40 min]: P=7.7 ×10
n=12; Fig. 3). Across datasets, network switching became in-
creasingly prominent with reduction of ALFF (diet: r=0.39, P=
0.00003; Leipzig: r=0.33, P=2.63 ×10
; Cam-CAN: r=0.25,
P=4.15 ×10
). Characterizing each network with respect to
its total ALFF-derived activity for all nodes, we then compared
symmetry for each switch: between transitions from lower- to
higher-activity networks versus transitions from higher- to lower-
activity networks. Both the ketogenic and fasting conditions showed
mean zero bias (one-sample ttest, keto diet: t=0.22, P=0.83; fast:
t=0.26, P=0.80), whereas the standard diet condition biased the
brain toward switching from higher- to lower-activity states
(standard diet: t=3.29, P=0.007). Thus, network switching may
reflect the brains inability to sustain the cost of more active,
metabolically taxing, networks, thereby defaulting to metabolically
cheaper(42) alternatives.
Our data provide evidence that, starting at around the age of 47 y,
the stability of brain networks begins to degrade with age, with the
most dramatic changes occurring around the age of 60 y. Since
glucose hypometabolism remains one of the hallmark clinical
features of dementia and its prodrome (43), we hypothesized that
the network destabilization seen with aging might reflect the
earliest stages of latent metabolic stress. Thus, we tested whether
diets with different energetic yield might modulate network sta-
bility even in a younger population expected to be decades prior to
any overt symptoms of age-based cognitive impairment. While
glucose is normally considered to be the brains default fuel,
β-hydroxybutyrate metabolism increases by 27% the Gibbs free
energy change for ATP compared to glucose (23, 24). Consistent
with that advantage, our results showed that even in younger
(<50 y) adults, dietary ketosis increased overall brain activity
and stabilized functional networks.
We first chose to manipulate diet in order to assess real-world
clinical implications of food choices on the brain. However, change
of diet within an ecologically realistic environment is a complex
variable and therefore cannot dissociate whether the observed
changes result from what is being taken away (carbohydrates) versus
what is being added (fat) or even whether the changes might reflect
different caloric intake (e.g., due to differences in satiety) for the
two conditions. We thus followed up with a second study in which
all participants followed their standard diets, fasted overnight, were
scanned in a fasted state, and were then scanned again 30 min after
drinking an individually weight-dosed and calorie-matched
bolus: glucose on one day and D-βHb ketone ester on the other,
counterbalanced for order. We found that the stabilizing effects seen
with dietary ketosis were replicated with administrat ion of exo ge-
nous ketones, which suggests that effects observed with mod-
ulating diet were specific to metabolism of glucose versus ketone
bodies rather than more holistic changes seen between diets.
It should be noted that one difficulty in isolating the impact of
each fuel type is the frequently observed (but potentially clini-
cally beneficial in its own right) side effect of exogenous ketones
in lowering glucose levels. This reflects a previously reported bias
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 navi-
gation 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 ttest 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).
| Mujica-Parodi et al.
Downloaded by guest on March 3, 2020
between the fuels: Ketone bodies, whenever present, are im-
mediately utilized by the brain regardless of need, whereas glu-
cose is only taken up by cells via GLUT transporters as required
(15, 44). Thus, in the (inherently physiologically unnatural) state
in which exogenous ketones are administered concomitantly with
glucose, ketone bodies saturate cells, and the cerebral metabolic
rate of glucose is down-regulated (44). However, ketone bodies
would stabilize networks by lowering glucose levels only if glucose
levels were already abnormally elevated, either due to insulin re-
sistance or in response to a physiological perturbative bolus. The
fact that network stabilizing effects were observed even in non-
insulin resistant individualstested in a stable state of dietary
glycolysissuggests that those effects were consequent to ketosis
rather than correcting a pathological state of hyperglycemia.
We next considered whether any systematic physiological ef-
fects of ketosis, such as diuresis (and therefore lowered blood
pressure) or reduced cellular need for oxygen, might confound
our fMRI results. However, if so, BOLD signal would have de-
creased in the ketogenic condition (45). The fact that ALFF and
network stability increased during that condition suggests that
the observed neurobiological effects did not result from global
changes in hydration or oxygen. On the other hand, since ketone
bodies have been shown to increase blood flow in the heart (46)
and brain (47), an increase in cerebral blood flow would be
consistent with increased BOLD, and therefore ALFF, but not
with the network behavior we observed. Experiments combining
arterial spin labeling and fMRI show increased cerebral blood
flow is associated with increased fMRI connectivity (48), a
modulation of connection strength. However, the observed, net-
work instability reflects a qualitatively different behavior, in which
networks transition between distinct topological configurations. We
believe this behavior is more consistent with mechanisms of syn-
aptic transmission, as suggested by previous animal experiments
(19). Establishing potential mechanisms by which energy avail-
ability, at the cellular level, affects reroutingof neural signals will
be an important future direction for multimodal and translational
For both diet and bolus experiments, D-βHb ketone ester and
fasting conditions produced equivalent effects in stabilizing brain
networks. Glycogen, when stored in the liver and skeletal muscle,
typically sustains glycolysis for fasts of up to 30 h. However, the
brain primarily utilizes glycogen stored in glia, which
has shown in humans to become depleted in 5 to 10 h (49).
Thus, following the typical overnight fast of 10 to 12 h, it is
likely that the brains of non-insulin-resistant participants had
already transitioned to endogenous ketosis, even if it was not yet
detectable with assays of peripheral ketosis measured by blood or
urine. Overall, our neuroimaging results support the hypothesis
that at least some of the beneficial neural effects reported with
hypocaloric states, such as intermittent fasting, severe caloric
restriction, and exercise, may result from the brains transition to
ketone bodies as fuel (50). While, for healthy individuals, the
benefits of endogenous ketosis may be naturally achieved in
multiple ways (e.g., ketogenic diet, fasting, exercise), this may not
be necessarily true for those with insulin resistance, as chroni-
cally elevated insulin levels associated with insulin resistance
present even during fasting (32)physiologically inhibit glucagon
and therefore ketogenesis (51). Thus, while we showed endoge-
nous and exogenous ketones to be qualitatively similar in stabi-
lizing brain networks in young healthy adults, for insulin-resistant
individuals, exogenous ketones may provide a useful adjunct in
achieving the neurobiological benefits seen with endogenous ke-
tosis, a further area for future study.
Finally, our focus on acute effects of modulating fuel source
controlled for the role of several potential mechanisms associ-
ated with differences seen in large-scale epidemiological studies
comparing diets. For example, insulin resistance has been
suggested to indirectly facilitate vascular dementia, as hyperglycemia
increases inflammation (52) and blocks nitric oxide (53), thereby
effectively narrowing brain vasculature while also increasing blood
viscosity (54). With respect to Alzheimers disease, recent results
(55) have identified an insulin-degrading enzyme as playing a
critical role in removing both excess insulin and amyloid β-protein
from the brain. Since insulin and the protein compete with one
another for the same enzyme, one consequence of the sustained
high insulin levels associated with insulin resistance is depletion of
the enzyme and therefore accumulated deposition of β-amyloid
plaque. In addition, ketones have been shown to reduce inflam-
mation and production of reactive oxygen species, as well as to up-
regulate mitochondria in the brain. While all of these may have
significant cumulative and synergistic effects in the months or
years that precede cognitive impairment, it is striking how quickly
the brain responded to a single week of dietary change or 30 min
following a single dose of D-βHb. This rapid response effectively
ruled out indirect inflammatory, antioxidant, tau/amyloid, and/or
adaptive mitochondrial mechanisms of action, allowing us to isolate
a more straightforward role of diet on metabolism. While further
experiments will be needed to elucidate the mechanism at a
microscopic scale and to explore its impact on the aging brain
over longer time periods, the near-immediate changes in network
stability, clearly visible even at the scale of the single participant,
are encouraging, as they suggest that dietary interventions can have
marked and measurable neurobiological effects on timescales
relevant to clinical intervention.
ACKNOWLEDGMENTS. The research described in this paper was funded by
the W. M. Keck Foundation (L.R.M.-P.), the White House Brain Research
Through Advancing Innovative Technologies (BRAIN) Initiative (Grants
NSFECCS1533257 and NSFNCS-FR 1926781 to L.R.M.-P.), and the US National
Academies (Grant NAKFICB8 to L.R.M.-P.). The authors gratefully acknowledge
assistance provided by Dominic DAgostino during the design phase and
Nathan A. Smith during the interpretation phase of the experiments.
1. D. D. Clark, L. Sokoloff, Circulation and energy metabolism of the brainin Basic
Neurochemistry: Molecular, Cellular and Medical Aspects, G. J. Siegel, B. W. Agranoff,
R.W.Albers,S.K.Risher,M.D.Uhler,Eds.(Lippincott, Philadelphia, 1999), pp. 637
2. L. Sokoloff, R. Mangold, R. L. Wechsler, C. Kenney, S. S. Kety, The effect of mental
arithmetic on cerebral circulation and metabolism. J. Clin. Invest. 34,11011108 (1955).
3. S. J. Olshansky et al., A potential decline in life expectancy in the United States in the
21st century. N. Engl. J. Med. 352, 11381145 (2005).
4. D. Dabelea et al.; SEARCH for Diabetes in Youth Study, Prevalence of type 1 and type
2 diabetes among children and adolescents from 2001 to 2009. JAMA 311, 17781786
5. M. Schnaider Beeri et al., Diabetes mellitus in midlife and the risk of dementia three
decades later. Neurology 63, 19021907 (2004).
6. R. H. Tuligenga et al., Midlife type 2 diabetes and poor glycaemic control as risk
factors for cognitive decline in early old age: A post-hoc analysis of the Whitehall II
cohort study. Lancet Diabetes Endocrinol. 2, 228235 (2014).
7. R. K. West et al., The association of duration of type 2 diabetes with cognitive per-
formance is modulated by long-term glycemic control. Am. J. Geriatr. Psychiatry 22,
10551059 (2014).
8. M. E. van Eersel et al., The interaction of age and type 2 diabetes on executive
function and memory in persons aged 35 years or older. PLoS One 8, e82991 (2013).
9. S. M. de la Monte, J. R. Wands, Alzheimers disease is type3 diabetes-evidence reviewed.
J. Diabetes Sci. Technol. 2,11011113 (2008).
10. J. M. Hill, M. A. Lesniak, C. B. Pert, J. Roth, Autoradiographic localization of insulin
receptors in rat brain: Prominence in olfactory and limbic areas. Neuroscience 17,
11271138 (1986).
11. L. D. Baker et al., Insulin resistance and Alzheimer-like reductions in regional cerebral
glucose metabolism for cognitively normal adults with prediabetes or early type 2
diabetes. Arch. Neurol. 68,5157 (2011).
12. S. Craft et al.,Memory improvement following induced hyperinsulinemia in Alzheimers
disease. Neurobiol. Aging 17, 123130 (1996).
13. H. Krebs, Biochemical aspects of ketosis. Proc. R. Soc. Med. 53,7180 (1960).
14. R. M. Sapolsky, Glucocorticoid toxicity in the hippocampus: Reversal by supplemen-
tation with brain fuels. J. Neurosci. 6, 22402244 (1986).
15. S. C. Cunnane et al., Can ketones help rescue brain fuel supply in later life? Implications
for cognitive health during aging and the treatment of Alzheimersdisease.Front. Mol.
Neurosci. 9,53(2016).
Mujica-Parodi et al. PNAS Latest Articles
Downloaded by guest on March 3, 2020
16. S. C. Cunnane et al., Can ketones compensate for deteriorating brain glucose uptake
during aging? Implications for the risk and treatment of Alzheimers disease. Ann. N.
Y. Acad. Sci. 1367,1220 (2016).
17. K. Eckles-Smith, D. Clayton, P. Bickford, M. D. Browning, Caloric restriction prevents
age-related deficits in LTP and in NMDA receptor expression. Brain Res. Mol. Brain
Res. 78, 154162 (2000).
18. N. Pitsikas, M. Carli, S. Fidecka, S. Algeri, Effect of life-long hypocaloric diet on age-
related changes in motor and cognitive behavior in a rat population. Neurobiol.
Aging 11, 417423 (1990).
19. S. Dutta, P. Sengupta, Men and mice: Relating their ages. Life Sci. 152, 244248 (2016).
20. M. Board et al., Acetoacetate is a more efficient energy-yielding substrate for human
mesenchymal stem cells than glucose and generates fewer reactive oxygen species.
Int. J. Biochem. Cell Biol. 88,7583 (2017).
21. P. Puchalska, P. A. Crawford, Multi-dimensional roles of ketone bodies in fuel me-
tabolism, signaling, and therapeutics. Cell Metab. 25, 262284 (2017).
22. J. C. Newman, E. Verdin, Ketone bodies as signaling metabolites. Trends Endocrinol.
Metab. 25,4252 (2014).
23. R. L. Veech, The therapeutic implications of ketone bodies: The effects of ketone
bodies inpathological conditions: Ketosis, ketogenic diet, redoxstates, insulin resistance,
and mitochondrial metabolism. Prostaglandins Leukot. Essent. Fatty Acids 70, 309319
24. K. Sato et al., Insulin, ketone bodies, and mitochondrial energy transduction. FASEB J.
9, 651658 (1995).
25. A. Fontán-Lozano et al., Caloric restriction increases learning consolidation and facili-
tates synaptic plasticity through mechanisms dependent on NR2B subunits of the NMDA
receptor. J. Neurosci. 27, 1018510195 (2007).
26. D. Attwell et al., Glial and neuronal control of brain blood flow. Nature 468, 232243
27. Y. Ma et al., Resting-state hemodynamics are spatiotemporally coupled to synchro-
nized and symmetric neural activity in excitatory neurons. Proc. Natl. Acad. Sci. U.S.A.
113, E8463E8471 (2016).
28. A. Babayan et al., A mind-brain-body dataset of MRI, EEG, cognition, emotion, and
peripheral physiology in young and old adults. Sci. Data 6, 180308 (2019).
29. J. R. Taylor et al., The Cambridge Centre for Ageing and Neuroscience (Cam-CAN)
data repository: Structural and functional MRI, MEG, and cognitive data from a cross-
sectional adult lifespan sample. Neuroimage 144, 262269 (2017).
30. D. J. DeDora et al., Signal fluctuation sensitivity: An improved metric for optimizing
detection of resting-state fMRI networks. Front. Neurosci. 10, 180 (2016).
31. N. Adra et al., Protecting the Aging Brain (PAgB), Project 1926781. Data Archive for
the Brain Initiative. Deposited 14 Feb-
ruary 2020.
32. D. R. Matthews et al., Homeostasis model assessment: Insulin resistance and beta-cell
function from fasting plasma glucose and insulin concentrations in man. Diabetologia
28, 412419 (1985).
33. V. Shivva et al., The population pharmacokinetics of D-β-hydroxybutyrate following
administration of (R)-3-Hydroxybutyl (R)-3-Hydr oxybutyrate. AAPS J. 18, 678688
34. World Health Organization, International Diabetes Foundation, Definition and di-
agnosis of diabetes mellitus and intermediate hyperglycaemia(WHO Press, Geneva,
Switzerland, 2006).
35. O. Esteban et al., fMRIPrep: A robust preprocessing pipeline for functional MRI. Nat.
Methods 16, 111116 (2019).
36. Y. Behzadi, K. Restom, J. Liau, T. T. Liu, A component based noise correction method
(CompCor) for BOLD and perfusion based fMRI. Neuroimage 37,90101 (2007).
37. W. R. Shirer, S. Ryali, E. Rykhlevskaia, V. Menon, M. D. Greicius, Decoding subject-driven
cognitive states with whole-brain connectivity patterns. Cereb. Cortex 22,158165
38. V. D. Blondel, J. L. Guillaume, R. Lambiotte, Fast unfolding of communities in large
networks. J. Stat. Mech. 2008, P10008 (2008).
39. G. P. Krishnan, O. C. González, M. Bazhenov, Origin of slow spontaneous resting-state
neuronal fluctuations in brain networks. Proc. Natl. Acad. Sci. U.S.A. 115, 68586863
40. D. Tomasi, N. D. Volkow, Association between brain activation and functional
connectivity. Cereb. Cortex 29, 19841996 (2019).
41. M. F. Folstein, S. E. Folstein, P. R. McHugh, Mini-mental state.A practical method
for grading the cognitive state of patients for the clinician. J. Psychiatr. Res. 12, 189
198 (1975).
42. A. Trevisiol et al., Monitoring ATP dynamics in electrically active white matter tracts.
eLife 6, e24241 (2017).
43. K. Chiotiset al., Longitudinal changes of tau PET imaging in relationto hypometabolism
in prodromal and Alzheimers disease dementia. Mol. Psychiatry 23,16661673 (2018).
44. A. Courchesne-Loyer et al., Inverse relationship between brain glucose and ketone
metabolism in adults during short-term moderate dietary ketosis: A dual tracer
quantitative positron emission tomography study. J. Cereb. Blo od Flow Metab. 37,
24852493 (2017).
45. R. Wang et al., Transient blood pressure changes affect the functional magnetic
resonance imaging detection of cerebral activation. Neuroimage 31,111 (2006).
46. L. C. Gormsen et al., Ketone body infusion with 3-hydroxybutyrate reduces myocardial
glucose uptake and increases blood flow in humans: A positron emission tomography
study. J. Am. Heart Assoc. 6, e005066 (2017).
47. S. G. Hasselbalch et al., Changes in cerebral blood flow and carbohydrate metabolism
during acute hyperketonemia. Am. J. Physiol. 270, E746E751 (1996).
48. M. Qiu, D. Scheinost, R. Ramani, R. T. Constable, Multi-modal analysis of functional
connectivity and cerebral blood flow reveals shared and unique effects of propofol in
large-scale brain networks. Neuroimage 148, 130140 (2017).
49. G. Oz et al., Human brain glycogen content and metabolism: Implications on its role
in brain energy metabolism. Am. J. Physiol. Endocrinol. Metab. 292, E946E951 (2007).
50. M. P. Mattson, K. Moehl, N. Ghena, M. Schmaedick, A. Cheng, Intermittent metabolic
switching, neuroplasticity and brain health. Nat. Rev. Neurosci. 19,6380 (2018).
51. K. G. Alberti, D. G. Johnston, A. Gill, A. J. Barnes, H. Orskov, Hormonal regulation of
ketone-body metabolism in man. Biochem. Soc. Symp., 163182 (1978).
52. M. Y. Donath, S. E. Shoelson, Type 2 diabetes as an inflammatory disease. Nat. Rev.
Immunol. 11,98107 (2011).
53. M. A. Creager, T. F. Lüscher, F. Cosentino, J. A. Beckman, Diabetes and vascular dis-
ease: Pathophysiology, clinical consequences, and medical therapy: Part I. Circulation
108,15271532 (2003).
54. R. B. Paisey, J. Harkness, M. Hartog, T. Chadwick, The effect of improvement in diabetic
control on plasma and whole blood viscosity. Diabetolog ia 19,345349 (1980).
55. W. Farris et al., Insulin-degrading enzyme regulates the levels of insulin, amyloid
beta-protein, and the beta-amyloid precursor protein intracellular domain in vivo.
Proc. Natl. Acad. Sci. U.S.A. 100, 41624167 (2003).
| Mujica-Parodi et al.
Downloaded by guest on March 3, 2020
... One mechanism for cognitive decline is reduced ability to utilize glucose for brain energy metabolism [2] and decreased brain network stability starting at age 47 [3], which may be circumvented by providing an alternative energy source such as ketones. A novel strategy to naturally improve metabolic health is to increase circulating ketones, namely beta-hydroxybutyrate (BHB), achievable though a ketogenic diet (KD) and/or exogenous ketones raising plasma ketone concentration levels into a range of nutritional ketosis (i.e., 0.5 to 4.0 mM) [4,5]. ...
... A novel strategy to naturally improve metabolic health is to increase circulating ketones, namely beta-hydroxybutyrate (BHB), achievable though a ketogenic diet (KD) and/or exogenous ketones raising plasma ketone concentration levels into a range of nutritional ketosis (i.e., 0.5 to 4.0 mM) [4,5]. Many studies have demonstrated improved cognitive performance in various neurological disorders using ketogenic therapies such as BHB infusion [6], ketone ester (KE) ingestion [3,7], and a KD [8,9]. ...
... The increase in BDNF after intense cycling exercise in Study 1 was not unexpected as previous studies have indicated an approximate 50% increase in BDNF after a maximal cycling protocol [28]. While acute exercise consistently increases BDNF, the response to long term training has shown either an increase [3,22,29,30] or no change in resting BDNF [31]. We demonstrated that a 12-week resistance training intervention resulted in no change in BDNF concentrations in the context of either a KD or MD. ...
Full-text available
Brain-Derived Neurotropic Factor (BDNF) expression is decreased in conditions associated with cognitive decline as well as metabolic diseases. One potential strategy to improve metabolic health and elevate BDNF is by increasing circulating ketones. Beta-Hydroxybutyrate (BHB) stimulates BDNF expression, but the association of circulating BHB and plasma BDNF in humans has not been widely studied. Here, we present results from three studies that evaluated how various methods of inducing ketosis influenced plasma BDNF in humans. Study 1 determined BDNF responses to a single bout of high-intensity cycling after ingestion of a dose of ketone salts in a group of healthy adults who were habitually consuming either a mixed diet or a ketogenic diet. Study 2 compared how a ketogenic diet versus a mixed diet impacts BDNF levels during a 12-week resistance training program in healthy adults. Study 3 examined the effects of a controlled hypocaloric ketogenic diet, with and without daily use of a ketone-salt, on BDNF levels in overweight/obese adults. We found that (1) fasting plasma BDNF concentrations were lower in keto-adapted versus non keto-adapted individuals, (2) intense cycling exercise was a strong stimulus to rapidly increase plasma BDNF independent of ketosis, and (3) clinically significant weight loss was a strong stimulus to decrease fasting plasma BDNF independent of diet composition or level of ketosis. These results highlight the plasticity of plasma BDNF in response to lifestyle factors but does not support a strong association with temporally matched BHB concentrations.
... Recently, exogenous sources of ketones, such as ketone esters, have been developed for their ability to elevate blood ketone concentrations without the need for changes in dietary macronutrient intake. These ketone esters have been used to test the effects of exogenous ketosis on a variety of end points across states of health and disease, ranging from physical (12)(13)(14)(15) and cognitive (16)(17)(18) performance to blood glucose regulation (19)(20)(21)(22) and cardiac function (23,24). The consumption of exogenous ketone products may replicate a subset of the effects of endogenous ketosis (6,25). ...
Full-text available
Objective: Growing interest in the metabolic state of ketosis has driven development of exogenous ketone products to induce ketosis without dietary changes. Bis hexanoyl (R)-1,3-butanediol (BH-BD) is a novel ketone ester which, when consumed, increases blood beta-hydroxybutyrate (BHB) concentrations. BH-BD is formulated as a powder or ready-to-drink (RTD) beverage; the relative efficacy of these formulations is unknown, but hypothesized to be equivalent. Methods: This randomized, observer-blinded, controlled, crossover decentralized study in healthy adults (n = 15, mean age = 33.7 years, mean BMI = 23.6 kg/m²) aimed to elucidate blood BHB and glucose concentrations before and 15, 30, 45, 60, 90 and 120 minutes following two serving sizes of reconstituted BH-BD powder (POW 25 g, POW 12.5 g), compared to a RTD BH-BD beverage (RTD 12.5 g), and a non-ketogenic control, all taken with a standard meal. Results: All BH-BD products were well tolerated and increased BHB, inducing nutritional ketosis (BHB ≥0.5 mM) after ∼15 minutes, relative to the control. BHB remained elevated 2 h post-consumption. The control did not increase BHB. Ketosis was dose responsive; peak BHB concentration and area under the curve (AUC) were two-fold greater with POW 25 g compared to POW 12.5 g and RTD 12.5 g. There were no differences in peak BHB and AUC between matched powder and RTD formulas. Blood glucose increased in all conditions following the meal but there were neither significant differences in lowest observed concentrations, nor consistent differences at each time point between conditions. These results demonstrate that both powdered and RTD BH-BD formulations similarly induce ketosis with no differences in glucose concentrations in healthy adults.
... By contrast, BHB treatment of hippocampal neurons reduced Aβ-mediated toxicity in vitro [212], as KD or KBs supplementation did in vivo in mouse models of AD [188,[197][198][199][200]. Additionally, KD or MCT supplementation increased INS signaling in the HPC of aged animals [196], specifically in neurons [213], and KD-fed adults showed lower levels of GLUT1, which transports glucose across the BBB, indicating that KBs decreased the dependence for glucose energetic metabolism [184]. It has been suggested that deleterious effects of INS resistance may result from metabolic stress, as neurons gradually lose access to glucose [214]. Therefore, KBs could provide neurons with an alternative fuel, reducing INS resistance-mediated brain damage and improving cognition in neurodegenerative diseases. ...
Full-text available
In recent decades, traditional eating habits have been replaced by a more globalized diet, rich in saturated fatty acids and simple sugars. Extensive evidence shows that these dietary factors contribute to cognitive health impairment as well as increase the incidence of metabolic diseases such as obesity and diabetes. However, how these nutrients modulate synaptic function and neuroplasticity is poorly understood. We review the Western, ketogenic, and paleolithic diets for their effects on cognition and correlations with synaptic changes, focusing mainly (but not exclusively) on animal model studies aimed at tracing molecular alterations that may contribute to impaired human cognition. We observe that memory and learning deficits mediated by high-fat/high-sugar diets, even over short exposure times, are associated with reduced arborization, widened synaptic cleft, narrowed post-synaptic zone, and decreased activity-dependent synaptic plasticity in the hippocampus, and also observe that these alterations correlate with deregulation of the AMPA-type glutamate ionotropic receptors (AMPARs) that are crucial to neuroplasticity. Furthermore, we explored which diet-mediated mechanisms modulate synaptic AMPARs and whether certain supplements or nutritional interventions could reverse deleterious effects, contributing to improved learning and memory in older people and patients with Alzheimer’s disease.
... It is an energy contributor for cellular activities [25,[29][30][31][32]. 3HB utilization is increased in atrophic cardiomyocyte [33,34], and exogenous 3HB exerts the obvious hemodynamic effects for patients with chronic heart failure (HF) [35,36]. The conventional ketogenic diet (KD) body supplement and 3HB supplemented to foods or drinks gradually have found applications for treating neurodegenerative disease such as epilepsy [37,38], Alzheimer's disease [39,40], cancer [41,42], aging [43], atherosclerosis [44], colonic inflammation and carcinogenesis [45], NLRP3-mediated inflammation [46], osteoporosis [47] and enhanced exercise performance [48]. ...
Full-text available
Background Muscle atrophy is an increasingly global health problem affecting millions, there is a lack of clinical drugs or effective therapy. Excessive loss of muscle mass is the typical characteristic of muscle atrophy, manifesting as muscle weakness accompanied by impaired metabolism of protein and nucleotide. (D)-3-hydroxybutyrate (3HB), one of the main components of the ketone body, has been reported to be effective for the obvious hemodynamic effects in atrophic cardiomyocytes and exerts beneficial metabolic reprogramming effects in healthy muscle. This study aims to exploit how the 3HB exerts therapeutic effects for treating muscle atrophy induced by hindlimb unloaded mice. Results Anabolism/catabolism balance of muscle protein was maintained with 3HB via the Akt/FoxO3a and the mTOR/4E-BP1 pathways; protein homeostasis of 3HB regulation includes pathways of ubiquitin–proteasomal, autophagic-lysosomal, responses of unfolded-proteins, heat shock and anti-oxidation. Metabolomic analysis revealed the effect of 3HB decreased purine degradation and reduced the uric acid in atrophied muscles; enhanced utilization from glutamine to glutamate also provides evidence for the promotion of 3HB during the synthesis of proteins and nucleotides. Conclusions 3HB significantly inhibits the loss of muscle weights, myofiber sizes and myofiber diameters in hindlimb unloaded mouse model; it facilitates positive balance of proteins and nucleotides with enhanced accumulation of glutamate and decreased uric acid in wasting muscles, revealing effectiveness for treating muscle atrophy. Graphical Abstract
Full-text available
Prior neuroimaging studies have shown associations between healthy lifestyle factors and cortical thickness; however, results on the direction of this association have been inconsistent. While the majority of studies were performed in older adults within specific weight status categories, little has been reported in younger populations with a range of adiposity, including groups with healthy-weight, overweight, and obesity. Here we investigated relationships between indices of physical activity (PA) and healthy eating with cortical thickness in children and youth/young adults and examined whether these relationships differed by weight status and age groups. Study participants included 119 youth/young adults and 159 children. We hypothesized that greater levels of PA and/or healthy eating index (HEI) composite scores would be positively associated with cortical thickness, and that this association would differ in overweight or obese groups versus normal weight groups, as well as youth/young adults vs. child cohorts. Overall PA (minutes/day) was assessed using 24-hour PA recalls. HEI was calculated to assess diet quality. A structural MRI was performed, and FreeSurfer 6.0 was used to assess cortical thickness in 68 regions of interest (ROI). Mixed effects modeling was performed to investigate associations of PA or HEI with cortical thickness. FDR corrections were applied for multiple ROIs. PA was positively associated with cortical thickness in the caudal middle frontal cortex (FDR adjusted p = 0.042) and cuneus cortex (FDR adjusted p = 0.017) after controlling for sex, age group, and weight status. When stratified by age, in youth/young adults, higher time spent in PA was associated with greater cortical thickness in the frontal, temporal, parietal and occipital cortex, after adjusting for sex and weight group (FDR adjusted ps < 0.05). No significant associations between PA and cortical thickness were observed in children. No significant associations between PA and cortical thickness were observed when stratified by weight group. No significant associations between HEI and cortical thickness were observed. These results indicate that higher time spent in PA is associated with greater cortical thickness, a relationship that appears to be stronger during youth/young adulthood and may be related to more favorable brain health outcomes.
Substance-use disorder (SUD) has become a global cause of morbidity and mortality and has an impact on general mental health. Substance users are mostly considered at high risk of nutritional deficiency, but most of the treatment centers do not provide any nutritional guide. The nutritional status of individuals in SUD needs to be properly addressed as it is responsible to cause malnutrition and making recovery more difficult. Hormonal alteration in SUD is the cause of low appetite and change in eating patterns, which leads to malnutrition. Furthermore, alterations in the absorption, utilization, metabolism, storage, distribution, and excretion of nutrients are responsible for nutritional deficiency. Micronutrient deficiency is reported in many of SUDs. Amino acid deficiency compromises the synthesis of neurotransmitters, which further exacerbates drug-seeking behavior. Moreover, craving for substances is altered by nutritional requirements, and the deprivation of food has shown to reduce the threshold for activation of reward pathways, which impedes recovery from SUD. During recovery, addiction transfer is a challenge in which patients start to crave for sweet food, which is also a cause of undernutrition and obesity. Hence, biopsychology of appetite and SUD can help to understand the causes of malnutrition. Therefore, in the recovery programs, planned nutrition can aid faster recovery and reduce the chances of food addiction and SUD relapse. Furthermore, alteration in dopaminergic and other brain pathways can also be considered via dietary intervention. It will not only be helpful in controlling the SUD, but also be beneficial for patient’s health.KeywordsSubstance-use disorderNutritionMalnutritionReward pathwayHormonesDietary intervention
The literature on large‐scale resting‐state functional brain networks across the adult lifespan was systematically reviewed. Studies published between 1986 and July 2021 were retrieved from PubMed. After reviewing 2938 records, 144 studies were included. Results on 11 network measures were summarized and assessed for certainty of the evidence using a modified GRADE method. The evidence provides high certainty that older adults display reduced within‐network and increased between‐network functional connectivity. Older adults also show lower segregation, modularity, efficiency and hub function, and decreased lateralization and a posterior to anterior shift at rest. Higher‐order functional networks reliably showed age differences, whereas primary sensory and motor networks showed more variable results. The inflection point for network changes is often the third or fourth decade of life. Age effects were found with moderate certainty for within‐ and between‐network altered patterns and speed of dynamic connectivity. Research on within‐subject bold variability and connectivity using glucose uptake provides low certainty of age differences but warrants further study. Taken together, these age‐related changes may contribute to the cognitive decline often seen in older adults. Although the literature on large‐scale, resting state functional networks in aging has been reviewed previously, we offer the first systematic qualitative and quantitative synthesis of the evidence. The novel synthesis stems from the adoption of PRISMA method and the breadth of network measures reviewed. The review offers a contemporary summary of the strength of the evidence, theoretical implications, and recommendations for further research.
Glucose metabolism is impaired in brain aging and several neurological conditions. Beneficial effects of ketones have been reported in the context of protecting the aging brain, however, their neurophysiological effect is still largely uncharacterized, hurdling their development as a valid therapeutic option. In this report, we investigate the neurochemical effect of the acute administration of a ketone d-beta-hydroxybutyrate (d-βHB) monoester in fasting healthy participants with ultrahigh-field proton magnetic resonance spectroscopy (MRS). In two within-subject metabolic intervention experiments, 7 T MRS data were obtained in fasting healthy participants (1) in the anterior cingulate cortex pre- and post-administration of d-βHB (N = 16), and (2) in the posterior cingulate cortex pre- and post-administration of d-βHB compared to active control glucose (N = 26). Effect of age and blood levels of d-βHB and glucose were used to further explore the effect of d-βHB and glucose on MRS metabolites. Results show that levels of GABA and Glu were significantly reduced in the anterior and posterior cortices after administration of d-βHB. Importantly, the effect was specific to d-βHB and not observed after administration of glucose. The magnitude of the effect on GABA and Glu was significantly predicted by older age and by elevation of blood levels of d-βHB. Together, our results show that administration of ketones acutely impacts main inhibitory and excitatory transmitters in the whole fasting cortex, compared to normal energy substrate glucose. Critically, such effects have an increased magnitude in older age, suggesting an increased sensitivity to ketones with brain aging.
Full-text available
Background and hypothesis: The robust evidence base supporting the therapeutic benefit of ketogenic diets in epilepsy and other neurological conditions suggests this same metabolic approach may also benefit psychiatric conditions. Study design: In this retrospective analysis of clinical care, 31 adults with severe, persistent mental illness (major depressive disorder, bipolar disorder, and schizoaffective disorder) whose symptoms were poorly controlled despite intensive psychiatric management were admitted to a psychiatric hospital and placed on a ketogenic diet restricted to a maximum of 20 grams of carbohydrate per day as an adjunct to conventional inpatient care. The duration of the intervention ranged from 6 to 248 days. Study results: Three patients were unable to adhere to the diet for >14 days and were excluded from the final analysis. Among included participants, means and standard deviations (SDs) improved for the Hamilton Depression Rating Scale scores from 25.4 (6.3) to 7.7 (4.2), P < 0.001 and the Montgomery-Åsberg Depression Rating Scale from 29.6 (7.8) to 10.1 (6.5), P < 0.001. Among the 10 patients with schizoaffective illness, mean (SD) of the Positive and Negative Syndrome Scale (PANSS) scores improved from 91.4 (15.3) to 49.3 (6.9), P < 0.001. Significant improvements were also observed in metabolic health measures including weight, blood pressure, blood glucose, and triglycerides. Conclusions: The administration of a ketogenic diet in this semi-controlled setting to patients with treatment-refractory mental illness was feasible, well-tolerated, and associated with significant and substantial improvements in depression and psychosis symptoms and multiple markers of metabolic health.
Full-text available
Objectives Exogenous ketone (monoester or salt) supplements are increasingly being employed for a variety of research purposes and marketed amongst the general public for their ability to raise blood beta-hydroxybutyrate (β-OHB). Emerging research suggests a blood glucose-lowering effect of exogenous ketones. Here, we systematically review and meta-analyze the available evidence of trials reporting on exogenous ketones and blood glucose. Methods We searched 6 electronic databases on December 13, 2021 for trials of any length that reported on the use of exogenous ketones compared to a placebo. We pooled raw mean differences (MD) in (i) blood β-OHB and (ii) blood glucose using random-effects models, and explored differences in the effects of ketone salts compared to ketone monoesters. Publication bias and risk of bias were examined using funnel plots and Cochrane's risk-of-bias tool, respectively. Results Twenty-eight trials including a total of 332 participants met inclusion criteria. There was no evidence for publication bias. Four trials were judged to be at low risk of bias with some concern for risk of bias in the remaining trials. Compared to placebo, consumption of exogenous ketones raised blood β-OHB (MD = 1.98 mM; 95% CI: 1.52 mM, 2.45 mM; P < 0.001) and decreased blood glucose (MD = −0.47 mM; 95% CI: −0.57 mM, −0.36 mM; P < 0.001) across the post-supplementation period of up to 300 minutes. Across both analyses, significantly greater effects were found following ingestion of ketone monoesters compared to ketone salts (P < 0.001). Conclusions Consumption of exogenous ketone supplements leads to acutely increased blood β-OHB and decreased blood glucose. Ketone monoesters exert a more potent β-OHB-raising and glucose-lowering effect as compared to ketone salts. Funding Sources Michael Smith Foundation for Health Research (MSFHR) Scholar Award.
Full-text available
We present a publicly available dataset of 227 healthy participants comprising a young (N=153, 25.1±3.1 years, range 20–35 years, 45 female) and an elderly group (N=74, 67.6±4.7 years, range 59–77 years, 37 female) acquired cross-sectionally in Leipzig, Germany, between 2013 and 2015 to study mind-body-emotion interactions. During a two-day assessment, participants completed MRI at 3 Tesla (resting-state fMRI, quantitative T1 (MP2RAGE), T2-weighted, FLAIR, SWI/QSM, DWI) and a 62-channel EEG experiment at rest. During task-free resting-state fMRI, cardiovascular measures (blood pressure, heart rate, pulse, respiration) were continuously acquired. Anthropometrics, blood samples, and urine drug tests were obtained. Psychiatric symptoms were identified with Standardized Clinical Interview for DSM IV (SCID-I), Hamilton Depression Scale, and Borderline Symptoms List. Psychological assessment comprised 6 cognitive tests as well as 21 questionnaires related to emotional behavior, personality traits and tendencies, eating behavior, and addictive behavior. We provide information on study design, methods, and details of the data. This dataset is part of the larger MPI Leipzig Mind-Brain-Body database.
Full-text available
Preprocessing of functional magnetic resonance imaging (fMRI) involves numerous steps to clean and standardize the data before statistical analysis. Generally, researchers create ad hoc preprocessing workflows for each dataset, building upon a large inventory of available tools. The complexity of these workflows has snowballed with rapid advances in acquisition and processing. We introduce fMRIPrep, an analysis-agnostic tool that addresses the challenge of robust and reproducible preprocessing for fMRI data. fMRIPrep automatically adapts a best-in-breed workflow to the idiosyncrasies of virtually any dataset, ensuring high-quality preprocessing without manual intervention. By introducing visual assessment checkpoints into an iterative integration framework for software testing, we show that fMRIPrep robustly produces high-quality results on a diverse fMRI data collection. Additionally, fMRIPrep introduces less uncontrolled spatial smoothness than observed with commonly used preprocessing tools. fMRIPrep equips neuroscientists with an easy-to-use and transparent preprocessing workflow, which can help ensure the validity of inference and the interpretability of results. © 2018, The Author(s), under exclusive licence to Springer Nature America, Inc.
Full-text available
The origin of the "resting-state" brain activity recorded with functional magnetic resonance imaging (fMRI) is still uncertain. Here we provide evidence for the neurovascular origins of the amplitude of the low-frequency fluctuations (ALFF) and the local functional connectivity density (lFCD) by comparing them with task-induced blood-oxygen level dependent (BOLD) responses, which are considered a proxy for neuronal activation. Using fMRI data for 2 different tasks (Relational and Social) collected by the Human Connectome Project in 426 healthy adults, we show that ALFF and lFCD have linear associations with the BOLD response. This association was significantly attenuated by a novel task signal regression (TSR) procedure, indicating that task performance enhances lFCD and ALFF in activated regions. We also show that lFCD predicts BOLD activation patterns, as was recently shown for other functional connectivity metrics, which corroborates that resting functional connectivity architecture impacts brain activation responses. Thus, our findings indicate a common source for BOLD responses, ALFF and lFCD, which is consistent with the neurovascular origin of local hemodynamic synchrony presumably reflecting coordinated fluctuations in neuronal activity. This study also supports the development of task-evoked functional connectivity density mapping.
Full-text available
During evolution, individuals whose brains and bodies functioned well in a fasted state were successful in acquiring food, enabling their survival and reproduction. With fasting and extended exercise, liver glycogen stores are depleted and ketones are produced from adipose-cell-derived fatty acids. This metabolic switch in cellular fuel source is accompanied by cellular and molecular adaptations of neural networks in the brain that enhance their functionality and bolster their resistance to stress, injury and disease. Here, we consider how intermittent metabolic switching, repeating cycles of a metabolic challenge that induces ketosis (fasting and/or exercise) followed by a recovery period (eating, resting and sleeping), may optimize brain function and resilience throughout the lifespan, with a focus on the neuronal circuits involved in cognition and mood. Such metabolic switching impacts multiple signalling pathways that promote neuroplasticity and resistance of the brain to injury and disease.
Full-text available
The development of tau-specific positron emission tomography (PET) tracers allows imaging in vivo the regional load of tau pathology in Alzheimer's disease (AD) and other tauopathies. Eighteen patients with baseline investigations enroled in a 17-month follow-up study, including 16 with AD (10 had mild cognitive impairment and a positive amyloid PET scan, that is, prodromal AD, and six had AD dementia) and two with corticobasal syndrome. All patients underwent PET scans with [(18)F]THK5317 (tau deposition) and [(18)F]FDG (glucose metabolism) at baseline and follow-up, neuropsychological assessment at baseline and follow-up and a scan with [(11)C]PIB (amyloid-β deposition) at baseline only. At a group level, patients with AD (prodromal or dementia) showed unchanged [(18)F]THK5317 retention over time, in contrast to significant decreases in [(18)F]FDG uptake in temporoparietal areas. The pattern of changes in [(18)F]THK5317 retention was heterogeneous across all patients, with qualitative differences both between the two AD groups (prodromal and dementia) and among individual patients. High [(18)F]THK5317 retention was significantly associated over time with low episodic memory encoding scores, while low [(18)F]FDG uptake was significantly associated over time with both low global cognition and episodic memory encoding scores. Both patients with corticobasal syndrome had a negative [(11)C]PIB scan, high [(18)F]THK5317 retention with a different regional distribution from that in AD, and a homogeneous pattern of increased [(18)F]THK5317 retention in the basal ganglia over time. These findings highlight the heterogeneous propagation of tau pathology among patients with symptomatic AD, in contrast to the homogeneous changes seen in glucose metabolism, which better tracked clinical progression.Molecular Psychiatry advance online publication, 16 May 2017; doi:10.1038/mp.2017.108.
Full-text available
Stem cells have been assumed to demonstrate a reliance on anaerobic energy generation, suited to their hypoxic in vivo environment. However, we found that human mesenchymal stem cells (hMSCs) have an active oxidative metabolism with a range of substrates. More ATP was consistently produced from substrate oxidation than glycolysis by cultured hMSCs. Strong substrate preferences were shown with the ketone body, acetoacetate, being oxidised at up to 35 times the rate of glucose. ROS-generation was 45-fold lower during acetoacetate oxidation compared with glucose and substrate preference may be an adaptation to reduce oxidative stress. The UCP2 inhibitor, genipin, increased ROS production with either acetoacetate or glucose by 2-fold, indicating a role for UCP2 in suppressing ROS production. Addition of pyruvate stimulated acetoacetate oxidation and this combination increased ATP production 27-fold, compared with glucose alone, which has implications for growth medium composition. Oxygen tension during culture affected metabolism by hMSCs. Between passages 2 and 5, rates of both glycolysis and substrate-oxidation increased at least 2-fold for normoxic (20% O2)- but not hypoxic (5% O2)-cultured hMSCs, despite declining growth rates and no detectable signs of differentiation. Culture of the cells with 3-hydroxybutyrate abolished the increased rates of these pathways. These findings have implications for stem cell therapy, which necessarily involves in vitro culture of cells, since low passage number normoxic cultured stem cells show metabolic adaptations without detectable changes in stem-like status.
Full-text available
Background High levels of ketone bodies are associated with improved survival as observed with regular exercise, caloric restriction, and—most recently—treatment with sodium–glucose linked transporter 2 inhibitor antidiabetic drugs. In heart failure, indices of ketone body metabolism are upregulated, which may improve energy efficiency and increase blood flow in skeletal muscle and the kidneys. Nevertheless, it is uncertain how ketone bodies affect myocardial glucose uptake and blood flow in humans. Our study was therefore designed to test whether ketone body administration in humans reduces myocardial glucose uptake (MGU) and increases myocardial blood flow. Methods and Results Eight healthy subjects, median aged 60 were randomly studied twice: (1) During 390 minutes infusion of Na‐3‐hydroxybutyrate (KETONE) or (2) during 390 minutes infusion of saline (SALINE), together with a concomitant low‐dose hyperinsulinemic–euglycemic clamp to inhibit endogenous ketogenesis. Myocardial blood flow was measured by ¹⁵O‐H2O positron emission tomography/computed tomography, myocardial fatty acid metabolism by ¹¹C‐palmitate positron emission tomography/computed tomography and MGU by ¹⁸F‐fluorodeoxyglucose positron emission tomography/computed tomography. Similar euglycemia, hyperinsulinemia, and suppressed free fatty acids levels were recorded on both study days; Na‐3‐hydroxybutyrate infusion increased circulating Na‐3‐hydroxybutyrate levels from zero to 3.8±0.5 mmol/L. MGU was halved by hyperketonemia (MGU [nmol/g per minute]: 304±97 [SALINE] versus 156±62 [KETONE], P<0.01), whereas no effects were observed on palmitate uptake oxidation or esterification. Hyperketonemia increased heart rate by ≈25% and myocardial blood flow by 75%. Conclusions Ketone bodies displace MGU and increase myocardial blood flow in healthy humans; these novel observations suggest that ketone bodies are important cardiac fuels and vasodilators, which may have therapeutic potentials.
Significance Resting-state infra-slow brain activity fluctuations are observed across various cognitive and disease brain states. Although resting-state fluctuations have received a great deal of interest over the past few years, the underlying biophysical mechanisms are not known. Using computational modeling, we show that spontaneous resting-state fluctuations arise from dynamic ion concentrations and are influenced by the Na ⁺ /K ⁺ pump, glial K ⁺ buffering, and AMPA/GABA synaptic currents. These findings provide insights into the biophysical mechanisms underlying generation of this phenomenon and may lead to better understanding of how different cognitive or disease states influence resting-state activity.
Synapses enable neurons to communicate with each other and are therefore a prerequisite for normal brain function. Presynaptically, this communication requires energy and generates large fluctuations in calcium concentrations. Mitochondria are optimized for supplying energy and buffering calcium, and they are actively recruited to presynapses. However, not all presynapses contain mitochondria; thus, how might synapses with and without mitochondria differ? Mitochondria are also increasingly recognized to serve additional functions at the presynapse. Here, we discuss the importance of presynaptic mitochondria in maintaining neuronal homeostasis and how dysfunctional presynaptic mitochondria might contribute to the development of disease.