[ 18 F]FDG functional PET revisited: A new perspective on the temporal dynamics of brain glucose metabolism

Abstract

[¹⁸F]FDG functional PET (fPET) enables investigation of dynamics in glucose metabolism occurring within seconds. However, the physiological mechanisms supporting rapid metabolic changes necessitate further attention to allow accurate interpretation. This work highlights candidate mechanisms driving [¹⁸F]FDG signal changes at high temporal resolution, offering complementary insights to existing interpretations. At rest, metabolic demands are closely matched by glucose supply across the blood-brain barrier (BBB), regulated by glucose transporter 1 (GLUT1). During neuronal activation, glucose transport and phosphorylation by hexokinase are elevated to meet increased energy requirements. Simulations indicate that rapid [¹⁸F]FDG signal increases are primarily driven by BBB transport, with subsequent increases in hexokinase activity. Mechanisms supporting increased BBB transport include elevated glucose concentration gradient towards the brain and changes in GLUT1 intrinsic properties, but only minor effects of blood flow. Conversely, moment-to-moment fluctuations in [¹⁸F]FDG used for metabolic connectivity, reflect temporally synchronized supply, mediated jointly by blood flow and BBB transport. We emphasize that the coupling between BBB transport and metabolism underpin the [¹⁸F]FDG fPET signal. Considering alterations of GLUT1 and subsequent metabolism in numerous brain disorders, stimulation-induced energy demands and metabolic connectivity represent a promising opportunity to investigate the underlying pathophysiological processes.

Keywords

Brain metabolism functional PET blood-brain barrier blood flow kinetic modeling

Introduction

[¹⁸F]FDG is the most commonly used radiotracer world-wide for positron emission tomography (PET).¹ This is due to its high clinical relevance in numerous medical fields² and the widespread availability given by the simple synthesis with an isotope half-life that allows transportation and commercial production.

Routine clinical use in brain imaging employs a bolus application, followed by a resting period of about 30–60 min and subsequent scanning for 10–15 min.³ Although this yields a static image, it is widely used as a surrogate for overall brain glucose metabolism, typically reported as the standard uptake value (SUV) or SUV ratio (SUVR). This interpretation is based on phosphorylation of the radiotracer and subsequent irreversible trapping in the cell (i.e., k₄ = 0 in a 2-tissue compartment model (2TCM), Figure 1). We acknowledge however that this assumption may not hold entirely for certain applications such as long scan times.

Figure 1.

Schematic description of the irreversible two-tissue compartment model, that is, k₄ = 0. Similar to glucose, the radiotracer [¹⁸F]FDG crosses the blood-brain barrier (BBB) via facilitated transport by the glucose transporter 1 (GLUT1, 55 kDa), described by rate constants k₁ and k₂. The uptake mechanism is in contrast to most other radiotracers, which cross the BBB via passive diffusion. Within the brain interstitium (unmetabolized compartment), glucose and [¹⁸F]FDG are taken up by neurons (GLUT3) and astrocytes (GLUT1, 45 kDA). Within the cells they are subsequently metabolized by hexokinase (k₃). The italic variables at the bottom refer to those of equation (3), being the concentration of glucose in plasma ([Gluc_Plasma]) and brain ([Gluc_Brain]), as well as the maximum (T_max) and actual transport rate of glucose (T_Gluc) across the BBB.

While a dynamic PET scan with a bolus application allows for identification of individual rate constants through compartment modeling of the radiotracer kinetics, it will in most cases still yield a static outcome image of the cerebral metabolic rate of glucose (CMRGlu). This can be derived from the influx constant K_i, which in turn is a macroparameter obtained from the individual rate constants:

K_{i} = K_{1} * k_{3} / (k_{2} + k_{3})

(1)

These rate constants describe the transport across the blood-brain barrier (BBB, K₁ and k₂), and the phosphorylation in the cell through the conversion of [¹⁸F]FDG to [¹⁸F]FDG-6-phosphate by the enzyme hexokinase (k₃), that is, the first step of glucose metabolism¹ (Figure 1). In most practical applications, SUV, SUVR, and CMRGlu images reflect brain metabolism at resting state, with k₃ showing the strongest correlation with these metrics at the above mentioned time scale.⁴ This offers the possibility to identify (non-acute) pathological alterations for example, caused by epilepsy, tumors, injuries as well as neurodegenerative and psychiatric disorders.

Further advancement of dynamic PET enables the quantification of different cognitive or emotional states within a single PET scan.^5
–7 Similar to functional magnetic resonance imaging (fMRI) this is achieved through presentation of for example, visual paradigms in a block design, thus yielding the term functional PET (fPET). The approach requires administration of radiotracers like [¹⁸F]FDG as a (bolus +) constant infusion to provide free radiotracer throughout the experiment, which is then able to be metabolized according to the current metabolic demands.^6,8 Task-related metabolism is commonly approximated by a ramp function,^6,7 implying a steady-state during the stimulation period. While this assumption is not strictly met, it provides a practical approximation^5,6 that has been shown to yield robust estimates when radiotracer uptake has stabilized.⁹ Another assumption of the approach is that the neuronal response and glucose metabolism are related in a linear and time-invariant manner, as in fMRI.¹⁰ This posits that changes in the neuronal response also translate to changes in metabolism with respect to amplitude and temporal coupling.^11,12

As the temporal resolution for this approach has substantially improved from PET image frames of 1 min^5,6 down to 3 s,¹³ acute adaptations in glucose metabolism become increasingly relevant. It is therefore essential to determine the assumptions and interpretations that apply to [¹⁸F]FDG fPET imaging at such a short time scale and to identify the possible neurophysiological mechanisms that underpin rapid changes in metabolic demands. This work aims to highlight major driving factors responsible for fast-acting changes in glucose metabolism and their links to the corresponding [¹⁸F]FDG signals measured with fPET in vivo. Although conceptual parallels are drawn to rate constants of the 2TCM, fPET is not assumed to unambiguously resolve their individual contribution to such rapid effects. Instead, this work uses simulations to explore plausible scenarios linking neurophysiological processes to observed signal dynamics, without implying uniqueness of the underlying parameter combinations. We will specifically focus on aspects applicable to short-term changes, which should be seen in addition to (not instead of) previous knowledge that explained glucose metabolism derived from steady-state kinetics obtained with conventional PET imaging. As such, the proposed concepts may in practice mostly apply to fPET imaging. This work follows a targeted approach in which key physiological mechanisms are synthesized from the literature and subsequently examined using analytical simulations of the irreversible 2TCM. The simulations are not intended as an independent methodological contribution, but rather serve to illustrate how different combinations of rate constants can reproduce experimentally observed [¹⁸F]FDG signal dynamics under physiologically plausible assumptions. Rather than providing a systematic review, we focus on mechanisms that are directly relevant to short-term changes in glucose metabolism at the temporal resolution of fPET. This knowledge will then be transferred to scenarios of acute stimulation-induced changes in [¹⁸F]FDG glucose metabolism. Finally, we propose that also moment-to-moment fluctuations in the fPET signal at resting state represent meaningful information. These enable computation of molecular connectivity, or more specifically metabolic connectivity when using [¹⁸F]FDG,^14
–16 but are subject to an even more refined interpretation. In sum, this work provides a neurophysiological account of the rapid dynamics in glucose metabolism elicited through cognitive or emotional processing as well as metabolic connectivity at resting-state, which can be imaged with [¹⁸F]FDG fPET.

Simulations

To characterize [¹⁸F]FDG uptake and metabolism induced by stimulation and moment-to-moment fluctuations, simulations were carried out using the irreversible 2TCM. The arterial input function and rate constants at resting-state were adopted from our previous work, using a typical bolus + infusion radiotracer administration protocol at a ratio of 20:80%,¹³ with K₁ = 0.11 ml/cm³/min, k₂ = 0.167 min⁻¹, k₃ = 0.074 min⁻¹, k₄ = 0 min⁻¹. These values are also in close agreement with previously published data.¹⁷ As outlined above, further key modeling assumptions include approximation of steady-state during task periods, linear and time-invariant coupling between neuronal activity and glucose metabolism as well as simplified representation of stimulation effects using ramp functions. The aim of the simulations is to explore which combinations of changes in rate constant can reproduce experimentally observed fPET signal characteristics, particularly the ramp-like response during stimulation and oscillatory fluctuations at rest. Given that true physiological changes in rate constants are not directly observable, the simulations are designed to identify plausible mechanistic scenarios rather than to estimate unique parameter sets. Accordingly, the simulations should be interpreted as illustrative and hypothesis-generating, rather than predictive or quantitatively validated.

Stimulation-specific changes in rate constants were introduced after 8 min to allow equilibration and lasted 36 s. Proceeding from our previous simplified simulation where changes in rate constants were stationary,¹³ the changes here were modeled as time-dependent variables. The model was implemented analytically in Matlab by solving the corresponding differential equations, while also considering the rate constants’ newly introduced time variance. Since the true changes in rate constants during neuronal activation are unknown, we explored which different sets of rate constants are able to characterize a ramp function during activation, similar to the one commonly used in fPET general linear model analyses.^6,8 For combined changes of rate constants (case 5 and 6 below) a minimization procedure was employed. The simulations were constrained to (i) reproduce experimentally observed increases in K_i ~27%¹³ (see below), and (ii) approximate the temporal profile of a ramp-like stimulation response. We would like to emphasize, that the presented parameter combinations are not unique. Multiple, potentially distinct, configurations of rate constant changes may produce similar [¹⁸F]FDG signal dynamics, limiting the ability to infer specific physiological mechanisms from the simulations alone. If possible, we assumed an increase in the individual constants’ values as this was deemed most likely from a physiological point of view. The following cases were explored:

∆K₁ = 5%

∆k₂ = 5%

∆k₃ = 5%

∆K₁: ∆k₂: ∆k₃ = x: −5x: 0, with x = 5%

∆K₁: ∆k₂: ∆k₃ = x: 0: 4.38x, with x = 6.81%

∆K₁: ∆k₂: ∆k₃ = x: 1.18x: 1.72x, with x = 19.89%

Cases 1–3 characterize changes in the [¹⁸F]FDG signal due to variation of individual rate constants alone. In case 4 only BBB-related transport rates are increased (i.e., ∆k₃ = 0). Note that from a mathematical view (but not an experimental one, see below), an increase in K₁ alone without changes in k₃ would be sufficient to cause an increase in overall metabolism K_i (see equation (1)). Case 5 assumes that neuronal activation leads only to increases in supply and metabolism of glucose and [¹⁸F]FDG with a constant reverse transport (i.e., ∆k₂ = 0). The rationale of case 6 is that BBB transport is generally increased maintaining a constant ratio between K₁ and k₂ (i.e., ∆K₁ = ∆k₂). To provide a more comprehensive picture of the effects of changes in rate constants, we also explored various other combinations (Supplemental Figure S1).

For metabolic connectivity, oscillating moment-to-moment fluctuations were simulated on top of basis radiotracer uptake (i.e., with K₁ = 0.11 ml/cm³/min, k₂ = 0.167 min⁻¹, k₃ = 0.074 min⁻¹, k₄ = 0 min⁻¹). That is, changes in rate constants as in case 6 above (i.e., with ∆K₁: ∆k₂: ∆k₃ = x: 1.18x: 1.72x) were repeatedly introduced for 30 s periods as observed in previous work,¹⁸ with the respective variations alternating between positive and negative algebraic signs. Furthermore, the contribution of k₃ was investigated by varying K₁ and k₂ only, with ∆k₃ = 0.

Finally, we indirectly assess the influence of blood flow, assuming that changes in CBF would lead to changes in the fractional volume of blood vB in the common operational equation used in PET imaging:

C_{P E T} = (1 - v B) * C_{T} + v B * C_{W B}

(2)

with C_PET, C_T, and C_WB representing the radiotracer concentration in the PET image, tissue and whole-blood, respectively. vB was changed according to the relationship established by Grubb et al.,¹⁹ yielding 8.85% change in CBV following increases in CBF by 25%.^11,20
–22 This was implemented for both, stimulation-induced changes (single stimulation block of 36 s) as well as resting-state fluctuations (repeatedly alternating changes of 30 s). These changes were modeled in analogy to BOLD signal changes, with a boxcar function for stimulation-induced changes and a sine-like change for resting-state fluctuations, respectively. In the simulations, only the vB part scaling C_WB was changed, as we assume that rapid CBF changes do not translate into corresponding changes in C_T on such a short time scale.

Basal glucose metabolism in the human brain

The human brain accounts for about 20% of the body’s energy demands under resting conditions, with glucose being the primary fuel. Since glucose is only scarcely stored in the brain (except for e.g., glycogen storage in astrocytes, see below), a constant supply from blood is mandatory to maintain brain function.^12,23 Glucose as well as [¹⁸F]FDG pass the BBB through facilitated transport via the glucose transporter 1 (GLUT1) carrier protein (although at different Michaelis–Menten constants K_M).^24
–26 The transport mechanism is required because of the hydrophilic nature of both compounds, which makes [¹⁸F]FDG different to most other radiotracers which cross the BBB via passive diffusion. Throughout the manuscript, GLUT1 refers to the form with a molecular mass of 55 kDa with highest expression at the BBB (Figure 1), while the 45 kDa form is expressed at astrocytes.^24,26

Glucose transport rate or velocity across the BBB (T_Gluc) can be described with the Michaelis–Menten equation, taking into account the difference in glucose concentration between blood plasma ([Gluc_Plasma]) and brain ([Gluc_Brain], equation (3)).²⁷ At rest, T_Gluc has been estimated to be about 37 µmol/100 g/min (i.e., 37% of its maximum value T_max = 1 µmol/g/min, see Dienel^12,27
–30 for assumptions).

\begin{array}{l} T_{Gluc} = T_{max} (\frac{1}{1 + (K_{M} / [{Gluc}_{Plasma}])} \\ - \frac{1}{1 + (K_{M} / [{Gluc}_{Brain}])}) \end{array}

(3)

with T_max being the maximal glucose transport rate across the BBB and K_M the Michaelis–Menten constant. This confirms the well-known fact that under normal resting conditions T_Gluc is not the limiting factor (rather, conversion by hexokinase is Furler et al.,³¹ Robinson and Rapoport³²), but it can become so during stimulation (see Leybaert,²⁷ Barros et al.,²⁹ Rothman et al.³³ and below).

As highlighted previously,²⁷ the interesting point is that the transport velocity of glucose across the BBB T_Gluc closely matches its metabolic rate CMRGlu at about 30 µmol/100 g/min.^6,12,27,34 In other words, at resting-state the delivery of glucose just satisfies the actual demands,^12,27,29 which is plausible given the limited storage in the form of glycogen^29,35 and the neurotoxicity of excessive glucose concentrations.³⁶ These facts in turn imply that acute increases in metabolic demands also require an increase in supply across the BBB as discussed in the next sections.

Stimulation-specific changes in [18F]FDG

We and others have repeatedly shown that stimulation-induced regional increases in CMRGlu as obtained with [¹⁸F]FDG in awake humans are stable at about 20%–25% from baseline metabolism across various experimental settings.^{6,27,33,37
–39} It is acknowledged that this is lower than the often referenced 50%,⁴⁰ but such a large change might be even beyond the maximum calculated transport rate of glucose.³³ Furthermore, recent work has reported increases in regional cerebral blood flow (CBF) of about 25%,^11,20
–22 that is, a similar range as compared to CMRGlu. Thus, the initially described coupling mechanisms between CMRGlu and CBF⁴⁰ still apply.

Simulation results for task effects

In line with studies that used other techniques than PET imaging,^38,41
–43 our recent work has demonstrated that such increases in [¹⁸F]FDG may occur almost instantly upon neuronal stimulation.¹³ To assess the contribution of individual rate constants and CBF, we carried out simulations of the 2TCM (Figure 2). Simulated changes in vB as presumably induced by CBF yielded instant step-wise changes in the [¹⁸F]FDG signal upon start and end of the stimulation, which did not resemble experimentally observed effects (Supplemental Figure S2; Figure 2(c)). Rather, initial increases in the [¹⁸F]FDG signal required changes in K₁ (and potentially k₂), that is, parameters that characterize BBB transport (Figure 2(a)). Furthermore, in the simulation a combination of changes in rate constants that also included changes in k₃ were required to match experimental data (Figure 2(b) and (c)). Thus, changes in k₃ were necessary for the [¹⁸F]FDG signal to return to baseline metabolism after the stimulus has ceased (i.e., slope = 0 in the task-specific time activity curve (TAC), Figure 2(b)), indicating that [¹⁸F]FDG is truly phosphorylated by hexokinase after transport into the brain. In line, previous work has elegantly shown that only an increase in both glucose transport across the BBB and its phosphorylation by hexokinase yields an increase in overall energy increase.^29,30

Figure 2.

Simulated and experimentally observed changes in the [¹⁸F]FDG signal upon neuronal activation. Simulations comprised changes in individual rate constants alone (dark blue, red, orange) as well as combinations thereof (green, purple, light blue). All time courses represent stimulation-specific effects, that is, after removing baseline radiotracer uptake, and were normalized by baseline radiotracer uptake. (a) Simulated increases in individual rate constants indicate that changes in K₁ (dark blue) and k₂ (red) lead to instant, near-linear changes in the fPET signal with subsequent return to baseline, whereas changes elicited by k₃ (orange) require more time to take effect. (b) Combinations of changes in K₁ and k₃ (purple) or K₁, k₂, and k₃ (light blue) were simulated to obtain a ramp function, that is typically used to model task-specific changes in glucose metabolism (dashed green line in (c)).⁶ Interestingly, changes in K₁ and k₂ alone (i.e., without changing k₃, green) resulted in a considerably different shape which is not compatible with experimental data, although such changes would mathematically be sufficient to yield changes in overall K_i. (c) Experimentally observed changes in the [¹⁸F]FDG fPET signal (black) show a rapid increase during the task, which returned to baseline when stimulation ceases.¹³ None of the individual rate constants (a) were able to explain the experimentally observed signal changes, but only their combination which included changes in k₃ (b) yielded a comparable effect. Task duration for simulations and experimental data was from 0 to 36 s (grey box). See section simulation for corresponding parameter settings.

This coupling of events may seem trivial, but it is particularly relevant for interpreting the rapid changes observed with fPET. To summarize the essential aspects underpinning the tight coupling of BBB transport and hexokinase activity upon neuronal activation, (i) at rest glucose supply matches the actual demands (T_Gluc–CMRGlu), (ii) there is almost no storage capacity for glucose in the brain, and (iii) independent simulations have shown that increases in BBB transport and hexokinase activity are both required to increase the actual metabolism. Again, these results imply that an acute surge in metabolic demands during neuronal activation must consequently also be matched by an increased supply. Previous work has already emphasized that glucose transport across the BBB represents an essential factor at rest⁴⁴ and during stimulation.^27,33,45 Of note, even earliest studies assessing task-specific glucose metabolism imaged BBB transport at least to some extent.^46,47 That is, task performance was carried out during the bolus injection of [¹⁸F]FDG, with the initial part of the TAC being mainly determined by K₁ and k₂.

Candidate mechanisms for BBB transport

Although glucose metabolism itself has been studied extensively, the corresponding supply has received relatively little attention. There are several candidate mechanisms available that may (or may not) explain such an increase in supply. Importantly, with respect to the [¹⁸F]FDG signal, all of these mechanisms will to some extent affect its transport via the BBB (K₁ and k₂) and subsequently its metabolism by hexokinase (k₃). These mechanisms likely occur in parallel, but it is currently not fully known, which of them is imperative and/or limiting.

Cerebral blood flow (CBF)

The increase in CBF upon neuronal activation is well established and represents the basis of blood-oxygen level dependent (BOLD) fMRI. Among others,⁴⁸ such changes occur at the level of the capillary bed by decreasing its resistance to and heterogeneity of flow, indicating a functional recruitment by increasing the fraction of high-flow capillaries.^49
–52 For PET imaging, changes in blood flow are often associated with K₁. This assumption is valid for radiotracers that cross the BBB via passive diffusion and have a high extraction fraction. However, as mentioned above glucose and [¹⁸F]FDG are subject to a facilitated transport mechanism via GLUT1. As such, glucose transport and metabolism are not particularly sensitive to variation in blood flow,⁵³ which is also reflected in the low permeability surface area product⁵⁴ and extraction fraction of about 10%^{12,27,30,55
–57} (Supplemental Figure S3). Experimental work using a hypercapnia challenge also showed that [¹⁸F]FDG uptake is not influenced by supraphysiological changes in CBF even if higher than 100%.⁵ Furthermore, moderate transient changes in blood glucose concentrations did not affect blood flow responses to stimulation,^58,59 indicating that CBF is not regulated in a metabolism-dependent feedback loop, but rather through neurotransmitter signaling.^60,61 This does not invalidate previous observations showing associations between CBF and CMRGlu. Rather, it suggests that these may be correlated, but only due to their common causal dependency on neuronal activation. Altogether, this indicates that changes in CBF are an accompanying but not sufficient condition for increased CMRGlu, thereby representing a minor contribution to changes in the [¹⁸F]FDG signal.

Glucose concentration gradient

The BBB is not freely permeable for glucose, maintaining a concentration gradient of about [Gluc_Plasma] = 5 mmol/L in blood plasma versus [Gluc_Brain] = 1 mmol/L in brain extracellular space at rest.^12,25,27 During neuronal activation, glucose phosphorylation is accelerated,²⁶ which decreases brain glucose concentration [Gluc_Brain] by about 20%.^{27,38,39,62,63} Leaving everything else constant in equation (3), such an increase in the concentration gradient between blood and brain alone would increase glucose transport across the BBB (T_Gluc) by 14%.²⁷ As noted previously, this effect derived from the law of mass action is still insufficient to explain CMRGlu changes in full.²⁸ Interestingly, the mentioned decrease in [Gluc_Brain] is very much within the range of regional changes observed with [¹⁸F]FDG fPET,^6,33,37 thereby providing the possible interpretation of fPET signal changes representing just the amount of glucose that is needed to satisfy the increased demands.

Intrinsic properties of GLUT1

Another aspect to modify BBB transport of glucose is given by changes in its permeability, which has also been described as neurobarrier coupling.²⁷ This is realized by adaptations of the intrinsic properties of GLUT1 upon activation, resulting in an increased T_max of equation (3).^64,65 Although rather little is known about the exact changes, potential mechanisms may include conformational changes between occluded and open states^66,67 as well as “unmasking” of already available binding sites.^64,68,69 Interestingly, the increases in T_max occurred without relevant changes in K_m.^64,70 Furthermore, GLUT1 appears in clusters, which are dissolved during activation resulting in increased glucose transport.⁷¹ Regarding the temporal scale, one study reported rapid changes in GLUT1 T_max within 10 s.⁷² However, this concerned the astrocytic GLUT1 transporter (45 kDa), which differs from that of the BBB (55 kDa), so it remains to be investigated whether these effects translate to BBB transport. In any case, slightly delayed increase in the transport is also in line with the above mentioned decrease in brain glucose concentration.³⁸ Unfortunately, quantitative estimations at short time scales are hardly available so far. One study demonstrated an increase in T_max by 30%–40%. However, this was obtained after the induction of seizures,⁷³ thus being a supraphysiological stimulus.

Number of GLUT1

Finally, a possibility to increase the transport rate of glucose is to increase the number of GLUT1 carrier proteins at the BBB surface. This likely occurs through redistribution of GLUT1 from the intracellular pool (which accounts for up to 40% of total GLUT1) to the membrane surface.^24,27 However, the time-scale at which this occurs has been described to be mostly within the range of minutes to hours.²⁷ Thus, increased expression of GLUT1 at the BBB may be relevant for subsequent adaptations to sustained neuronal activation, but probably represents a minor factor in the adaptations to meet acute metabolic demands. Also, the above changes in T_max may even occur without variations in the number of glucose transporters⁶⁴ and mRNA.⁷⁴

In sum, these findings suggest that the majority of increases in glucose supply upon neuronal activation are related to an increased concentration gradient towards the brain, followed by changes of the intrinsic properties of GLUT1, whereas increases in blood flow only play a minor role. Thus, the increased supply reflects the fast increase of the [¹⁸F]FDG signal, which is then followed by its metabolism via hexokinase.

Resting-state dynamics of [18F]FDG

The assessment of resting-state functional connectivity as derived from BOLD fMRI and EEG^75,76 has made a convincing case that spontaneous fluctuations of brain signals represent biologically meaningful information even in the absence of specific stimulation. This has substantially changed our understanding about the interaction between brain regions, organized in hierarchical networks each underpinning different physiological functions.^77
–79

Although the concept to compute connectivity with PET data was introduced decades ago,^80,81 the computation of molecular connectivity has received particular attention in recent years,^14,16,82 due to the ability to obtain reliable high temporal resolution PET data. In analogy to BOLD fMRI functional connectivity, molecular connectivity refers to the correlation of fluctuations of signal time courses between brain regions.¹⁴ It has been shown that the computation of metabolic connectivity with [¹⁸F]FDG data also features structured network organization.^15,16 However, the interpretation and the underlying neurophysiological mechanisms of such rapid signal fluctuations in the fPET signal without external stimulation have not been established.

Moment-to-moment fluctuations in [¹⁸F]FDG

There are various approaches available to compute moment-to-moment fluctuations from [¹⁸F]FDG fPET data,^{9,14
–16,83} but all with the common aim to remove the cumulative part of the TAC reflecting the irreversible uptake of the radiotracer. The most illustrative approach is a bandpass filter, due to the straightforward interpretation in terms of the signals’ frequencies.

Simulation results for metabolic connectivity

Simulations were carried out to assess potential contributions to the observed signal changes. Simulated changes in individual rate constants were first Fourier transformed, with focus on the frequency range of 0.01 < f < 0.1 Hz,¹⁵ commonly used for BOLD fMRI functional connectivity. The corresponding frequency spectra indicate that the [¹⁸F]FDG signal underlying these fluctuations is primarily determined by K₁ and k₂ reflecting transport across the BBB, rather than k₃ (Figure 3(a)), as the effect of changes in k₃ was simply too slow (Figures 2(a) and 3(a)). Next, combined changes in rate constants were simulated and subsequently band-pass filtered in the 0.01–0.1 Hz frequency range. This procedure effectively removed the irreversible radioligand uptake, leaving only oscillating changes in the signal, both in experimental and simulated data (Figure 3(b) and (c)). Notably, the resulting oscillations resulting from the simulations were almost identical, regardless of whether changes in k₃ changes were included. However and in contrast to task-induced changes, simulated variations in vB (reflecting presumed CBF changes) lead to similar oscillations in the [¹⁸F]FDG signal (Figure 3(b)).

Figure 3.

Simulated and experimentally observed changes in the [¹⁸F]FDG signal at resting state. (a) Frequency spectra of stimulation-specific changes in the TACs after simulated increases in individual rate constants as shown in Figure 2(a), (i.e., after removing baseline radiotracer uptake). The excerpt (lower panel) indicates that changes in K₁ (dark blue) and to some extent k₂ (red) comprise frequency components beyond 0.01 Hz, but not k₃ (orange). (b) Combined changes in K₁, k₂, and k₃ (light blue) exhibit oscillating changes on top of a steady baseline uptake. Rate constants were changed at the same ratio as in Figure 2(b; light blue). After applying a bandpass filter these oscillating changes remain (lower panel). Of note, omitting changes in k₃ in this simulation showed virtually identical results, indicating that k₃ does not contribute to these changes. Such oscillatory fluctuations were also obtained when varying the blood volume component vB in the 2TCM (orange), indicating that CBF indeed plays a relevant role for resting-state metabolic fluctuations. However, the exact ratio of contributions between changes in rate constants and CBF still remains unknown and as such the amplitudes may change. (c) Experimentally observed changes in the [¹⁸F]FDG fPET signal of a representative subject.⁸⁴ Application of a bandpass filter leaves oscillating changes also for experimental data (lower panel), which represent the basis for metabolic connectivity.

Several aspects however warrant further investigation and experimental validation. For instance, it remains to be clarified to what extent CBF variations translate into vB changes at the capillary level. Furthermore, the employed 2TCM has been developed for modeling tracer kinetics in steady-state, whereas rapid fluctuations may require additional refinements. Consequently, the precise contributions of individual rate constants and CBF remains uncertain and the amplitudes of the effects shown in Figure 3(b) may change under different modeling assumptions. These findings should therefore be interpreted cautiously, as the relative contributions of transport, phosphorylation, and blood flow may vary substantially depending on the underlying modeling assumptions. Nevertheless, it seems unlikely that metabolic connectivity obtained with [¹⁸F]FDG exclusively reflects CBF-related effects, since independent studies have reported substantial differences between BOLD-derived functional connectivity and [¹⁸F]FDG-based metabolic connectivity in terms of overall network structure^15,16 and alterations in aging.⁸⁵

Potential interpretation of metabolic connectivity

From a more general view, it seems plausible that the observed fluctuations reflect adaptations through a regulatory circuit, aiming to maintain a constant supply of glucose which matches the current consumption, that is, slight increases and decreases from a steady baseline according to the actual changes in demand. In the hypothetical case that every brain region has equal metabolic demands, this oscillation would be uniform across the entire brain. However, since the brain is never truly at rest⁸⁶ and different brain regions are involved in processing distinct stimuli, its metabolic demands also vary across space and time, resulting in a regional variability in this regulatory signal. If two brain regions have equal metabolic demands simultaneously, then also their time courses would be similar and thus correlated, irrespective of their baseline metabolism. These regions can be considered to be metabolically coupled through their common acute changes in metabolic demands,^80,81 which are in turn matched by corresponding changes in CBF and BBB transport. Recent work on high-temporal resolution fPET data showed a distinct network-specific organization of metabolic connectivity as computed from [¹⁸F]FDG signal fluctuations.^15,18 However, the exact neurobiological underpinnings still need to be established. Although it might be surprising that k₃ does not seem to play a direct role in metabolic connectivity, it is of course still relevant. As emphasized throughout this manuscript and within the concept of neurobarrier coupling,²⁷ changes in BBB transport are initially triggered by neuronal activation, subsequently leading to increased glucose consumption and thus increases in k₃ (see above).

Limitations, outlook, and conclusions

Numerous studies with simulations and experimental data demonstrate that stimulation-induced increases in glucose metabolism require BBB transport for subsequent phosphorylation by hexokinase. We emphasize that changes in BBB transport may not be seen independently from hexokinase activity, as neuronal activation triggers both in order to match energy supply and demand, that is, these mechanisms are strongly coupled. In terms of [¹⁸F]FDG fPET imaging this implies that several rate constants (especially K₁ and k₃) are involved in this process (see Supplemental Figure S4 for association between rate constants). The adaptations supporting neuronal activation are mainly provided by an increased glucose BBB gradient towards the brain and functional changes in the intrinsic properties of GLUT1 molecules, while blood flow plays only a minor role for stimulation-specific effects. Conversely, moment-to-moment fluctuations in the [¹⁸F]FDG signal used for metabolic connectivity seem to reflect the regulation of metabolic supply through CBF and across the BBB to match the underlying demands.

As a limitation, current models may not yet be sufficient to characterize such rapid changes in glucose metabolism and the corresponding [¹⁸F]FDG signal. That is, the Michaelis–Menten equation is valid for the steady-state, an assumption that does not hold for acute changes in neuronal activation. Thus, several variables (e.g., T_max, [Gluc_Brain]) become time-variant, which requires a different mathematical solution. For the simulations of the 2TCM, time-dependent rate constants were used, but the changes were only assumed to occur during the task effect at a fixed rate, while temporally varying changes and those beyond the stimulation period are possible. A further limitation is the non-identifiability of the model, as multiple combinations of time-varying rate constants can produce indistinguishable [¹⁸F]FDG TACs. Although our examples cannot cover the full simulation space, we provided results within a physiologically plausible and thus representative range of changes. Still, a full validation of the model against experimental data remains to be conducted in future work. Finally, we acknowledge that other sources apart from glucose may also be used for energy production. For instance, glycogen is stored in astrocytes, but its turnover is slower than that of glucose.³⁰ Furthermore, recent modelling suggested that glycogenolysis takes place in astrocytes to spare glucose for neuronal energy metabolism upon activation.^33,35 This model also seemed to explain fluctuations in lactate,⁸⁷ which may however not represent a major energy source under normal conditions.⁸⁸ Nevertheless, the current work outlines potential neurophysiological underpinnings of rapid changes in the [¹⁸F]FDG signal, It is important to note that the described mechanisms do not replace previous assumptions and interpretations of [¹⁸F]FDG imaging, rather they should be seen as additional considerations specifically relevant for short-term changes imaged with fPET.

Future work may build upon this information to investigate if known pathological alterations of brain glucose metabolism can be characterized in more detail. For instance, decreases in GLUT1 transporters at the BBB explain hypometabolism, which occurs before the onset of cognitive dysfunction and amplifies neurodegenerative processes,^89
–91 alterations that are also found in Parkinson’s disease.^55,92 Further, increased GLUT1 phosphorylation in major depressive disorder was alleviated in patients who remitted after treatment.⁹³ Thus, BBB transfer by GLUT1 may represent a promising therapeutic target for various disorders to rescue brain glucose metabolism.^25,94

Finally, the herein proposed concepts may at least partly translate to other targets such as dopamine and serotonin synthesis, imaged with 6-[¹⁸F]FDOPA⁹⁵ and [¹¹C]AMT fPET,⁹⁶ respectively, as numerous of the presented aspects apply in a similar manner. These include facilitated transport across the BBB (via the L-type amino acid transporter 1),^97,98 a low extraction fraction from blood⁹⁹ as well as tight coupling of BBB transport¹⁰⁰ and neurotransmitter synthesis enzymes^101
–104 with neuronal activation. Future application of this approach may therefore provide novel insights into neurotransmitter dysfunction in psychiatric disorders.

Supplemental Material

sj-docx-1-jcb-10.1177_0271678X261455750 – Supplemental material for [18F]FDG functional PET revisited: A new perspective on the temporal dynamics of brain glucose metabolism

Supplemental material, sj-docx-1-jcb-10.1177_0271678X261455750 for [18F]FDG functional PET revisited: A new perspective on the temporal dynamics of brain glucose metabolism by Andreas Hahn, Pia Falb, Matej Murgaš, Sebastian Klug, Murray B. Reed, Godber M. Godbersen, Magdalena Ponce de León, Christian Milz, Marcus Hacker, Dan Rujescu and Rupert Lanzenberger in Journal of Cerebral Blood Flow & Metabolism

Footnotes

Acknowledgements

We thank the graduated team members and the diploma students of the Neuroimaging Lab (NIL, head: RL) as well as the clinical colleagues from the Department of Psychiatry and Psychotherapy for clinical and/or administrative support. In detail, we would like to thank L. Silberbauer, and G. Gryglewski for medical support, L. Rischka for acquisition and analysis support. The scientific project was performed with the support of the Medical Imaging Cluster of the Medical University of Vienna.

Author contributions

Conceptualization: AH and RL. Methodology: PF, AH, MBR, and MM. Software: PF. Writing – original draft: AH, MM, and PF. Writing – review and editing: All authors. Funding acquisition: AH, RL, and CM. All authors discussed the implications, revised the manuscript and approved the final version.

Declaration of conflicting interests

The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: RL received investigator-initiated research funding from Siemens Healthcare regarding clinical research using PET/MR. He is a shareholder of the start-up company BM Health GmbH since 2019. DR served as consultant for Janssen, received honoraria from Gerot Lannacher, Janssen and Pharmagenetix, received travel support from Angelini and Janssen, and served on advisory boards of AC Immune, Roche and Rovi. M. Hacker received consulting fees and/or honoraria from Bayer Healthcare BMS, Eli Lilly, EZAG, GE Healthcare, Ipsen, ITM, Janssen, Roche, and Siemens Healthineers. All other authors report no conflict of interest in relation to this study.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This research was funded in whole or in part by the Austrian Science Fund (FWF; Grant DOI 10.55776/KLI610, 10.55776/KLI1151 and 10.55776/PAT5436523, PI: A. Hahn, Grant DOI 10.55776/KLI1006, PI: RL). For open access purposes, the author has applied a CC BY public copyright license to any author-accepted manuscript version arising from this submission. CM is a recipient of a DOC Fellowship (27221) from the Austrian Academy of Sciences at the Department of Psychiatry and Psychotherapy, Medical University of Vienna.

ORCID iDs

Andreas Hahn

Pia Falb

Sebastian Klug

Murray B. Reed

Magdalena Ponce de León

Christian Milz

Rupert Lanzenberger

Data and code availability

Code for simulations is available at Github:

Supplemental material

Supplemental material for this article is available online.

References

Wadsak

Mitterhauser

Basics and principles of radiopharmaceuticals for PET/CT. Eur J Radiol 2010; 73(3): 461–469.

Buerkle

Weber

WA.

Imaging of tumor glucose utilization with positron emission tomography. Cancer Metastasis Rev 2008; 27(4): 545–554.

Guedj

Varrone

Boellaard

, et al EANM procedure guidelines for brain PET imaging using [(18)F]FDG, version 3. Eur J Nucl Med Mol Imaging 2022; 49(2): 632–651.

Volpi

Lee

Vlassenko

, et al The brain’s “dark energy” puzzle upgraded: [(18)F]FDG uptake, delivery and phosphorylation, and their coupling with resting-state brain activity. J Cereb Blood Flow Metab 2025; 45(9): 1799–1815.

Villien

Wey

Mandeville

, et al Dynamic functional imaging of brain glucose utilization using fPET-FDG. Neuroimage 2014; 100: 192–199.

Hahn

Gryglewski

Nics

, et al Quantification of task-specific glucose metabolism with constant infusion of 18F-FDG. J Nucl Med 2016; 57(12): 1933–1940.

Jamadar

Ward

, et al Simultaneous task-based BOLD-fMRI and [18-F] FDG functional PET for measurement of neuronal metabolism in the human visual cortex. Neuroimage 2019; 189: 258–266.

Rischka

Gryglewski

Pfaff

, et al Reduced task durations in functional PET imaging with [(18)F]FDG approaching that of functional MRI. Neuroimage 2018; 181: 323–330.

Hahn

Reed

Milz

, et al A unified approach for identifying PET-based neuronal activation and molecular connectivity with the functional PET toolbox. J Cereb Blood Flow Metab 2025; 46(2): 558–568.

10.

Poldrack

Mumford

Nichols

TE.

Handbook of functional MRI data analysis. New York, NY: Cambridge University Press, 2011.

11.

Kessler

RM.

Imaging methods for evaluating brain function in man. Neurobiol Aging 2003; 24(Suppl 1): S21–S35; discussion S37–S39.

12.

Dienel

GA.

Brain glucose metabolism: integration of energetics with function. Physiol Rev 2019; 99(1): 949–1045.

13.

Hahn

Reed

Vraka

, et al High-temporal resolution functional PET/MRI reveals coupling between human metabolic and hemodynamic brain response. Eur J Nucl Med Mol Imaging 2023; 51(5): 1310–1322.

14.

Reed

Cocchi

Sander

, et al Connecting the dots: approaching a standardized nomenclature for molecular connectivity in positron emission tomography. Eur J Nucl Med Mol Imaging 2025; 53(1): 48–58.

15.

Reed

Graf

Murgaš

, et al High-temporal resolution metabolic connectivity resolved by component-based noise correction. J Cereb Blood Flow Metab. Epub ahead of print 2026. DOI: 10.1177/0271678X261431043.

16.

Jamadar

Ward

PGD

Liang

, et al Metabolic and hemodynamic resting-state connectivity of the human brain: a high-temporal resolution simultaneous BOLD-fMRI and FDG-fPET multimodality study. Cereb Cortex 2021; 31(6): 2855–2867.

17.

Reivich

Alavi

Wolf

, et al Glucose metabolic rate kinetic model parameter determination in humans: the lumped constants and rate constants for [18F]fluorodeoxyglucose and [11C]deoxyglucose. J Cereb Blood Flow Metab 1985; 5(2): 179–192.

18.

Reed

Graf

Murgaš

, et al A comparative evaluation of molecular connectivity and covariance approaches. BioRxiv 2025: 2025.12.17.694867. DOI: 10.64898/2025.12.17.694867.

19.

Grubb

Jr Raichle

Eichling

, et al The effects of changes in PaCO2 on cerebral blood volume, blood flow, and vascular mean transit time. Stroke 1974; 5(5): 630–639.

20.

Fang

Ohba

Hashimoto

, et al Imaging mitochondrial complex I activation during a vibrotactile stimulation: a PET study using [(18)F]BCPP-EF in the conscious monkey brain. J Cereb Blood Flow Metab 2020; 40(12): 2521–2532.

21.

Mintun

Lundstrom

Snyder

, et al Blood flow and oxygen delivery to human brain during functional activity: theoretical modeling and experimental data. Proc Natl Acad Sci U S A 2001; 98(12): 6859–6864.

22.

Fox

Raichle

ME.

Focal physiological uncoupling of cerebral blood flow and oxidative metabolism during somatosensory stimulation in human subjects. Proc Natl Acad Sci U S A 1986; 83(4): 1140–1144.

23.

Jamadar

Behler

Deery

, et al The metabolic costs of cognition. Trends Cogn Sci 2025; 29(6): 541–555.

24.

Simpson

Carruthers

Vannucci

SJ.

Supply and demand in cerebral energy metabolism: the role of nutrient transporters. J Cereb Blood Flow Metab 2007; 27(11): 1766–1791.

25.

Patching

SG.

Glucose transporters at the blood-brain barrier: function, regulation and gateways for drug delivery. Mol Neurobiol 2017; 54(2): 1046–1077.

26.

Koepsell

Glucose transporters in brain in health and disease. Pflügers Archiv Eur J Physiol 2020; 472(9): 1299–1343.

27.

Leybaert

Neurobarrier coupling in the brain: a partner of neurovascular and neurometabolic coupling?

J Cereb Blood Flow Metab 2005; 25(1): 2–16.

28.

Leybaert

De Bock

Van Moorhem

, et al Neurobarrier coupling in the brain: adjusting glucose entry with demand. J Neurosci Res 2007; 85(15): 3213–3220.

29.

Barros

Porras

Bittner

CX.

Why glucose transport in the brain matters for PET. Trends Neurosci 2005; 28(3): 117–119.

30.

Barros

Bittner

Loaiza

, et al A quantitative overview of glucose dynamics in the gliovascular unit. Glia 2007; 55(12): 1222–1237.

31.

Furler

Jenkins

Storlien

, et al In vivo location of the rate-limiting step of hexose uptake in muscle and brain tissue of rats. Am J Physiol 1991; 261(3 Pt 1): E337–E347.

32.

Robinson

Rapoport

SI.

Glucose transport and metabolism in the brain. Am J Physiol 1986; 250(1 Pt 2): R127–R136.

33.

Rothman

Dienel

Behar

, et al Glucose sparing by glycogenolysis (GSG) determines the relationship between brain metabolism and neurotransmission. J Cereb Blood Flow Metab 2022; 42(5): 844–860.

34.

Gruetter

Novotny

Boulware

, et al 1H NMR studies of glucose transport in the human brain. J Cereb Blood Flow Metab 1996; 16(3): 427–438.

35.

Dienel

Cruz

NF.

Contributions of glycogen to astrocytic energetics during brain activation. Metab Brain Dis 2015; 30(1): 281–298.

36.

Tomlinson

Gardiner

NJ.

Glucose neurotoxicity. Nat Rev Neurosci 2008; 9(1): 36–45.

37.

Godbersen

Falb

Klug

, et al Non-invasive assessment of stimulation-specific changes in cerebral glucose metabolism with functional PET. Eur J Nucl Med Mol Imaging 2024; 51(8): 2283–2292.

38.

Wilson

GS.

Rapid changes in local extracellular rat brain glucose observed with an in vivo glucose sensor. J Neurochem 1997; 68(4): 1745–1752.

39.

Chen

Novotny

Zhu

, et al Localized 1H NMR measurement of glucose consumption in the human brain during visual stimulation. Proc Natl Acad Sci U S A 1993; 90(21): 9896–900.

40.

Fox

Raichle

Mintun

, et al Nonoxidative glucose consumption during focal physiologic neural activity. Science 1988; 241(4864): 462–464.

41.

Freeman

RD.

Neurometabolic coupling between neural activity, glucose, and lactate in activated visual cortex. J Neurochem 2015; 135(4): 742–754.

42.

Roche

Salazar

Martin

, et al Simultaneous measurements of glucose, oxyhemoglobin and deoxyhemoglobin in exposed rat cortex. J Neurosci Methods 2011; 202(2): 192–198.

43.

Mann

Deny

Ganguli

, et al Coupling of activity, metabolism and behaviour across the Drosophila brain. Nature 2021; 593(7858): 244–248.

44.

Hawkins

Mans

Davis

, et al Glucose availability to individual cerebral structures is correlated to glucose metabolism. J Neurochem 1983; 40(4): 1013–1018.

45.

Hargreaves

Planas

Cremer

, et al Studies on the relationship between cerebral glucose transport and phosphorylation using 2-deoxyglucose. J Cereb Blood Flow Metab 1986; 6(6): 708–716.

46.

Phelps

Kuhl

Mazziota

JC.

Metabolic mapping of the brain’s response to visual stimulation: studies in humans. Science 1981; 211(4489): 1445–1448.

47.

Vlassenko

Rundle

Mintun

MA.

Human brain glucose metabolism may evolve during activation: findings from a modified FDG PET paradigm. Neuroimage 2006; 33(4): 1036–1041.

48.

Mishra

Reynolds

Chen

, et al Astrocytes mediate neurovascular signaling to capillary pericytes but not to arterioles. Nat Neurosci 2016; 19(12): 1619–1627.

49.

Kleinfeld

Mitra

Helmchen

, et al Fluctuations and stimulus-induced changes in blood flow observed in individual capillaries in layers 2 through 4 of rat neocortex. Proc Natl Acad Sci 1998; 95(26): 15741–15746.

50.

Vogel

Kuschinsky

Decreased heterogeneity of capillary plasma flow in the rat Whisker-Barrel cortex during functional hyperemia. J Cereb Blood Flow Metab 1996; 16(6): 1300–1306.

51.

Akgören

Lauritzen

Functional recruitment of red blood cells to rat brain microcirculation accompanying increased neuronal activity in cerebellar cortex. NeuroReport 1999; 10(16): 3257.

52.

Barfod

Akgoren

Fabricius

, et al Laser-Doppler measurements of concentration and velocity of moving blood cells in rat cerebral circulation. Acta Physiol Scand 1997; 160(2): 123–132.

53.

Gejl

Rungby

Brock

, et al At the centennial of Michaelis and Menten, competing Michaelis-Menten steps explain effect of GLP-1 on blood-brain transfer and metabolism of glucose. Basic Clin Pharmacol Toxicol 2014; 115(2): 162–171.

54.

Chung

Abdelhafez

Spencer

, et al Quantitative PET imaging and modeling of molecular blood-brain barrier permeability. Nat Commun 2025; 16(1): 3076.

55.

Barros

San Martín

Ruminot

, et al Near-critical GLUT1 and neurodegeneration. J Neurosci Res 2017; 95(11): 2267–2274.

56.

Hasselbalch

Holm

Pedersen

, et al The (18)F-fluorodeoxyglucose lumped constant determined in human brain from extraction fractions of (18)F-fluorodeoxyglucose and glucose. J Cereb Blood Flow Metab 2001; 21(8): 995–1002.

57.

Duffy

Caldwell

Hoiland

, et al Brain glucose extraction is fixed at 10% despite twofold variability in resting cerebral blood flow in healthy humans. J Cereb Blood Flow Metab 2026; 46(3): 663–670.

58.

Powers

Hirsch

Cryer

PE.

Effect of stepped hypoglycemia on regional cerebral blood flow response to physiological brain activation. Am J Physiol 1996; 270(2 Pt 2): H554–H559.

59.

Wolf

Lindauer

Villringer

, et al Excessive oxygen or glucose supply does not alter the blood flow response to somatosensory stimulation or spreading depression in rats. Brain Res 1997; 761(2): 290–299.

60.

Attwell

Buchan

Charpak

, et al Glial and neuronal control of brain blood flow. Nature 2010; 468(7321): 232–243.

61.

Mergenthaler

Lindauer

Dienel

, et al Sugar for the brain: the role of glucose in physiological and pathological brain function. Trends Neurosci 2013; 36(10): 587–597.

62.

Fellows

Boutelle

Fillenz

Extracellular brain glucose levels reflect local neuronal activity: a microdialysis study in awake, freely moving rats. J Neurochem 1992; 59(6): 2141–2147.

63.

McNay

Fries

Gold

PE.

Decreases in rat extracellular hippocampal glucose concentration associated with cognitive demand during a spatial task. Proc Natl Acad Sci U S A 2000; 97(6): 2881–2885.

64.

Barnes

Ingram

Bennett

MDM

, et al Methyl-beta-cyclodextrin stimulates glucose uptake in Clone 9 cells: a possible role for lipid rafts. Biochem J 2004; 378(2): 343–351.

65.

Caliceti

Zambonin

Prata

, et al Effect of plasma membrane cholesterol depletion on glucose transport regulation in leukemia cells. PLoS One 2012; 7(7): e41246.

66.

Park

M-S.

Molecular dynamics simulations of the human glucose transporter GLUT1. PLoS One 2015; 10(4): e0125361.

67.

Mitrovic

McComas

Alleva

, et al Reconstructing the transport cycle in the sugar porter superfamily using coevolution-powered machine learning. eLife 2023; 12: e84805.

68.

Hamrahian

Zhang

J-Z

Elkhairi

, et al Activation of Glut1 glucose transporter in response to inhibition of oxidative phosphorylation: role of sites of mitochondrial inhibition and mechanism of Glut1 activation. Arch Biochem Biophys 1999; 368(2): 375–379.

69.

Barnes

Ingram

Porras

, et al Activation of GLUT1 by metabolic and osmotic stress: potential involvement of AMP-activated protein kinase (AMPK). J Cell Sci 2002; 115(11): 2433–2442.

70.

Mercado

Loeb

Ismail-Beigi

Enhanced glucose transport in response to inhibition of respiration in Clone 9 cells. Am J Physiol Cell Physiol 1989; 257(1): C19–C28.

71.

Yan

Zhou

, et al Mechanistic insights into GLUT1 activation and clustering revealed by super-resolution imaging. Proc Natl Acad Sci 2018; 115(27): 7033–7038.

72.

Loaiza

Porras

Barros

LF.

Glutamate triggers rapid glucose transport stimulation in astrocytes as evidenced by real-time confocal microscopy. J Neurosci 2003; 23(19): 7337–7342.

73.

Cornford

Nguyen

Landaw

EM.

Acute upregulation of blood-brain barrier glucose transporter activity in seizures. Am J Physiol Heart Circ Physiol 2000; 279(3): H1346–H1354.

74.

Shetty

Loeb

Vikstrom

, et al Rapid activation of GLUT-1 glucose transporter following inhibition of oxidative phosphorylation in clone 9 cells. J Biol Chem 1993; 268(23): 17225–17232.

75.

Biswal

Yetkin

Haughton

, et al Functional connectivity in the motor cortex of resting human brain using echo-planar MRI. Magn Reson Med 1995; 34(4): 537–541.

76.

Fox

Snyder

Vincent

, et al The human brain is intrinsically organized into dynamic, anticorrelated functional networks. Proc Natl Acad Sci U S A 2005; 102(27): 9673–9678.

77.

Fox

Raichle

ME.

Spontaneous fluctuations in brain activity observed with functional magnetic resonance imaging. Nat Rev Neurosci 2007; 8(9): 700–711.

78.

Margulies

Ghosh

Goulas

, et al Situating the default-mode network along a principal gradient of macroscale cortical organization. Proc Natl Acad Sci U S A 2016; 113(44): 12574–12579.

79.

Smith

Fox

Miller

, et al Correspondence of the brain’s functional architecture during activation and rest. Proc Natl Acad Sci U S A 2009; 106(31): 13040–13045.

80.

Horwitz

Duara

Rapoport

SI.

Intercorrelations of glucose metabolic rates between brain regions: application to healthy males in a state of reduced sensory input. J Cereb Blood Flow Metab 1984; 4(4): 484–499.

81.

McIntosh

Grady

Ungerleider

, et al Network analysis of cortical visual pathways mapped with PET. J Neurosci 1994; 14(2): 655–666.

82.

Sala

Lizarraga

Caminiti

, et al

Brain connectomics: time for a molecular imaging perspective?

Trends Cogn Sci 2023; 27(4): 353–366.

83.

Reed

Ponce

Leon

Vraka

, et al Whole-body metabolic connectivity framework with functional PET. Neuroimage 2023; 271: 120030.

84.

Silberbauer

Reed

Gryglewski

, et al Acute effects of selective serotonin reuptake inhibitors on cerebral glucose metabolism and blood flow. Transl Psychiatry 2025; 16(1): 54.

85.

Deery

Liang

Moran

, et al Metabolic connectivity has greater predictive utility for age and cognition than functional connectivity. Brain Commun 2025; 7(1): fcaf075.

86.

Raichle

ME.

The restless brain. Brain Connect 2011; 1(1): 3–12.

87.

Wilson

GS.

A temporary local energy pool coupled to neuronal activity: fluctuations of extracellular lactate levels in rat brain monitored with rapid-response enzyme-based sensor. J Neurochem 1997; 69(4): 1484–1490.

88.

Dienel

GA.

Lack of appropriate stoichiometry: strong evidence against an energetically important astrocyte–neuron lactate shuttle in brain. J Neurosci Res 2017; 95(11): 2103–2125.

89.

Winkler

Nishida

Sagare

, et al GLUT1 reductions exacerbate Alzheimer’s disease vasculo-neuronal dysfunction and degeneration. Nat Neurosci 2015; 18(4): 521–530.

90.

Erickson

Banks

WA.

Blood–brain barrier dysfunction as a cause and consequence of Alzheimer’s disease. J Cereb Blood Flow Metab 2013; 33(10): 1500–1513.

91.

Szablewski

Glucose transporters in brain: in health and in Alzheimer’s disease. J Alzheimer’s Dis 2017; 55(4): 1307–1320.

92.

Zhang

Lachance

Mattson

, et al Glucose metabolic crosstalk and regulation in brain function and diseases. Prog Neurobiol 2021; 204: 102089.

93.

Kahl

Georgi

Bleich

, et al Altered DNA methylation of glucose transporter 1 and glucose transporter 4 in patients with major depressive disorder. J Psychiatr Res 2016; 76: 66–73.

94.

Cunnane

Trushina

Morland

, et al Brain energy rescue: an emerging therapeutic concept for neurodegenerative disorders of ageing. Nat Rev Drug Discov 2020; 19(9): 609–633.

95.

Hahn

Reed

Pichler

, et al Functional dynamics of dopamine synthesis during monetary reward and punishment processing. J Cereb Blood Flow Metab 2021; 41(11): 2973–2985.

96.

Hahn

Reed

Murgas

, et al Dynamics of human serotonin synthesis differentially link to reward anticipation and feedback. Mol Psychiatry 2024; 30(2): 600–607.

97.

del Amo

Urtti

Yliperttula

Pharmacokinetic role of L-type amino acid transporters LAT1 and LAT2. Eur J Pharm Sci 2008; 35(3): 161–174.

98.

Zaragoza

Transport of amino acids across the blood-brain barrier. Front Physiol 2020; 11: 973.

99.

Muzik

Chugani

Chakraborty

, et al Analysis of [C-11]alpha-methyl-tryptophan kinetics for the estimation of serotonin synthesis rate in vivo. J Cereb Blood Flow Metab 1997; 17(6): 659–669.

100.

Delescluse

Simonnet

Ziegler

, et al A LAT1-like amino acid transporter regulates neuronal activity in the drosophila mushroom bodies. Cells 2024; 13(16): 1340.

101.

Hery

Simonnet

Bourgoin

, et al Effect of nerve activity on the in vivo release of [3H]serotonin continuously formed from L-[3H]tryptophan in the caudate nucleus of the cat. Brain Res 1979; 169(2): 317–334.

102.

Hamon

Bourgoin

Artaud

, et al The role of intraneuronal 5-HT and of tryptophan hydroxylase activation in the control of 5-HT synthesis in rat brain slices incubated in K+-enriched medium. J Neurochem 1979; 33(5): 1031–1042.

103.

Morgenroth

3rd Boadle-Biber

Roth

RH.

Tyrosine hydroxylase: activation by nerve stimulation. Proc Natl Acad Sci U S A 1974; 71(11): 4283–4287.

104.

Neff

Hadjiconstantinou

Aromatic L-amino acid decarboxylase modulation and Parkinson’s disease. Prog Brain Res 1995; 106: 91–97.

Supplementary Material

Please find the following supplemental material available below.

For Open Access articles published under a Creative Commons License, all supplemental material carries the same license as the article it is associated with.

For non-Open Access articles published, all supplemental material carries a non-exclusive license, and permission requests for re-use of supplemental material or any part of supplemental material shall be sent directly to the copyright owner as specified in the copyright notice associated with the article.

1.30 MB

0.00 MB

[ 18 F]FDG functional PET revisited: A new perspective on the temporal dynamics of brain glucose metabolism

Abstract

Keywords

Introduction

Simulations

Basal glucose metabolism in the human brain

Stimulation-specific changes in [18F]FDG

Simulation results for task effects

Candidate mechanisms for BBB transport

Cerebral blood flow (CBF)

Glucose concentration gradient

Intrinsic properties of GLUT1

Number of GLUT1

Resting-state dynamics of [18F]FDG

Moment-to-moment fluctuations in [18F]FDG

Simulation results for metabolic connectivity

Potential interpretation of metabolic connectivity

Limitations, outlook, and conclusions

Supplemental Material

sj-docx-1-jcb-10.1177_0271678X261455750 – Supplemental material for [18F]FDG functional PET revisited: A new perspective on the temporal dynamics of brain glucose metabolism

Footnotes

Acknowledgements

Author contributions

Declaration of conflicting interests

Funding

ORCID iDs

Data and code availability

Supplemental material

References

Supplementary Material

Moment-to-moment fluctuations in [¹⁸F]FDG