Brain connectivity changes to fast versus slow dopamine increases

The rewarding effects of stimulant drugs such as methylphenidate (MP) depend crucially on how fast they raise dopamine in the brain. Yet how the rate of drug-induced dopamine increases impacts brain network communication remains unresolved. We manipulated route of MP administration to generate fast versus slow dopamine increases. We hypothesized that fast versus slow dopamine increases would result in a differential pattern of global brain connectivity (GBC) in association with regional levels of dopamine D1 receptors, which are critical for drug reward. Twenty healthy adults received MP intravenously (0.5 mg/kg; fast dopamine increases) and orally (60 mg; slow dopamine increases) during simultaneous [11C]raclopride PET-fMRI scans (double-blind, placebo-controlled). We tested how GBC was temporally associated with slow and fast dopamine increases on a minute-to-minute basis. Connectivity patterns were strikingly different for slow versus fast dopamine increases, and whole-brain spatial patterns were negatively correlated with one another (rho = −0.54, pspin < 0.001). GBC showed “fast>slow” associations in dorsal prefrontal cortex, insula, posterior thalamus and brainstem, caudate and precuneus; and “slow>fast” associations in ventral striatum, orbitofrontal cortex, and frontopolar cortex (pFDR < 0.05). “Fast>slow” GBC patterns showed significant spatial correspondence with D1 receptor availability (estimated via normative maps of [11C]SCH23390 binding; rho = 0.22, pspin < 0.05). Further, hippocampal GBC to fast dopamine increases was significantly negatively correlated with self-reported ‘high’ ratings to intravenous MP across individuals (r(19) = −0.68, pbonferroni = 0.015). Different routes of MP administration produce divergent patterns of brain connectivity. Fast dopamine increases are uniquely associated with connectivity patterns that have relevance for the subjective experience of drug reward.

The SUVr method for estimating dynamic dopamine increases can be seen as an approximation of LSSRM (see demonstration below).LSSRM requires only one scan session with an MP challenge to estimate dynamic dopamine increases, but it necessitates five fit parameters, which hindered reliable quantification of dopamine in our data.This may be in part because we were unable to continuously infuse [ 11 C]raclopride throughout the 90 minutes of scanning (this was due to the challenges posed by the simultaneous PET-MRI setup.Specifically, to ensure safety, our magnetic pump had to be positioned six feet away from the MRI bore.Consequently, utilizing the bolus-plus-infusion method for [ 11 C]raclopride would have necessitated excessively high levels of initial radioactivity (>80mCi), which was deemed unsafe.)Therefore, radioactivity counts were lower at the end of the scan than in a paradigm with a continuous infusion.While LSSRM does not strictly require a paradigm with a continuous infusion, in our dataset we found that the relatively low radioactivity counts made dopamine quantification with LSSRM challenging.However, our design had the advantage of an additional placebo scan for each participant.Therefore, we developed an approach that capitalized on the added reliability the placebo scan affords, and could overcome the lack of a continuous [ 11 C]raclopride infusion.While the ΔSUVr approach requires two scans (MP and placebo) it has an important advantage: it only requires the amplitude of ΔSUVr and the time-to-peak of its derivative for fitting the ΔSUVr data, which improved the reliability of dynamic 'dopamine increases' estimates over prior methods.
The Simplified Reference Tissue Model (SRTM) defines the kinetic  !() of a target region in relation to the kinetic  " () of a reference region 1 .
represents the local rate of delivery in the target tissue compared to the reference tissue, with  $ representing the transfer rate constant from tissue to blood in the reference region, and  $' representing the transfer rate constant from tissue to blood in the target region.The linear extension of the simplified reference region model (LSSRM 2 ) extended this model by incorporating a time-varying efflux rate  $' () =  $' +  ℎ() that accounts for the competition between the radioligand and the endogenous neurotransmitter at the receptor sites.Here  represents the magnitude of transient effects and the function ℎ() characterizes the endogenous neurotransmitter discharge or an exogenous concurrent drug concentration level.Since MP increases extracellular dopamine, it also increases binding competition and reduces tracer concentration in the target region, Eq [1] can be expressed as: The standardized uptake value, (), is calculated by dividing the uptake value in a specific region of interest (ROI) by the uptake value in a reference region.The reference region is typically an area of the brain that is considered to have minimal specific binding for the radiotracer used in the PET study.The  is used as a simplified way to quantify the relative accumulation or binding of a radiotracer in a particular brain region compared to the reference region.
The  is beneficial because it allows for comparison and analysis of PET data across different individuals or studies by normalizing the values to a reference region.This normalization accounts for potential variations in overall radiotracer uptake due to factors such as individual differences in blood flow or metabolism.
The  change, Δ(), caused by MP-related increases in endogenous dopamine quantifies the change in radiotracer binding with respect to the placebo condition.
The ΔSUVr approach offers a significant advantage over prior methods, such as LSSRM, by eliminating the need for individual-specific SRTM parameters (R1, k2, k2a) to estimate dopamine increases.This enhances the robustness of model fitting as it only requires the amplitude of ΔSUVr and the time-to-peak of its derivative for fitting the ΔSUVr data. (8) Simulations demonstrating the similarity between the Δ method used in the current study to estimate dynamic dopamine increases, and the 'linear simplified reference region model' (LSSRM) method that has been used in prior studies.A) Time-varying concentrations of [11C]raclopride in the striatum,  ) (), and in the cerebellum,  * ().B) A gamma variate function modeling the endogenous dopamine increases elicited by methylphenidate (MP), ℎ().C) Dynamics of SUVr changes, Δ(), caused by MP-related increases in endogenous dopamine modeled with the LSSRM (black) and the approximation used in this study.D) Linear association between the exact and approximated LSSRM solutions.Normal random noise (5%) was added to  * () and  ) ().Dashed lines indicate the time of MP injection.

Figure S2 .
Figure S2.Brain connectivity changes to slow and fast dopamine increases show largely opposing patterns.Group maps (onesample t-tests) depicting the fit between global brain connectivity (GBC) and fast (intravenous methylphenidate; top) or slow (oral methylphenidate; bottom) maps.These results are the same as presented in Figure2Aof the main manuscript; here we are showing more detailed maps overlaid on coronal brain slices.We also highlight several notable regions, in purple circles, where connectivity patterns diverged: Precuneus, Insula, Nucleus Accumbens (NAcc) and dorsolateral prefrontal cortex (dlPFC).

Figure S3 .
Figure S3.Placebo control analysis: brain connectivity changes to slow and fast dopamine increases show largely opposing patterns.Left column: Group maps (one-sample t-tests) depicting the fit between global brain connectivity (GBC) and slow (oral methylphenidate (MP); top) or fast (IV MP; bottom) dopamine (DA) increase maps (identical to Figure2Aof main manuscript).Middle column: additional control analyses directly comparing the speed of DA increases in the active drug conditions to that of the placebo condition (i.e., paired t-tests representing IV MP > Placebo and Oral MP > Placebo).These results were highly similar to the original analyses (Supplementary FigureS2) demonstrating specificity of the findings to the active drug conditions.Right column: The GBC maps to oral MP > placebo versus IV MP > placebo were significantly negatively correlated.Purple dots represent the null distribution (controlling for spatial autocorrelation); red dot is the observed correlation (rho = -.60;pspin < .001).Thus, the primary finding of opposing connectivity patterns to slow and fast DA increases in the main manuscript (Figure2B) was preserved when controlling for placebo effects.Color bars represent t-values.

Figure S4 .
Figure S4.Dynamic global brain connectivity (GBC) changes with dopamine rate and association with the cortical distribution of dopamine D1 receptors.A) Left: direct comparison (paired t-test) of GBC associations with fast versus slow dopamine increases to methylphenidate.Right: normative PET map of D1 receptor density (using [ 11 C]SCH23390 data from the neuromaps repository).B) Spatial permutation test demonstrating a significant positive spatial correlation between "fast > slow" GBC associations and D1 receptor density.Purple dots represent the null distribution (controlling for spatial autocorrelation); red dot is the observed correlation.