Neural circuit selective for fast but not slow dopamine increases in drug reward

The study addresses how the speed of drug-induced dopamine increases influences human brain circuits underlying subjective drug reward. Faster routes of administration (e.g., intravenous or smoking) produce stronger rewarding effects and higher addiction risk than slower routes (e.g., oral), despite similar total dopamine transporter blockade. Prior human PET work showed that oral and IV methylphenidate (MP) can achieve similar overall striatal dopamine increases, but only fast IV delivery reliably produces a 'high'. Animal work indicates faster striatal dopamine increases and stronger engagement of reward circuitry with rapid delivery, and epidemiology links fast delivery methods to more severe substance use disorders. Beyond striatum, dopamine modulates large-scale cortical networks, including ventromedial prefrontal cortex (vmPFC) and salience network regions (dorsal anterior cingulate cortex, insula) implicated in reward, craving, and addiction remission after lesions. The authors hypothesized that slow dopamine increases (primarily D2-mediated) would decrease cortical activity, whereas fast increases (additionally engaging low-affinity D1 receptors) would elicit increases in activity, leading to distinct cortical-striatal activation and connectivity patterns for oral versus IV MP that relate to subjective 'high'.

- Human and animal studies indicate that speed of drug delivery determines intensity of reward and addictive potential, with IV and smoked routes producing faster dopamine rises and greater reward than oral or intranasal routes. - Early PET studies found that while oral and IV MP can yield comparable overall striatal dopamine increases, only IV MP reliably induces 'high' in healthy adults; cocaine users report highest 'high' with smoked, then IV, then intranasal routes at matched transporter blockade. - Reward processing extends beyond striatum to cortical circuits: vmPFC and nucleus accumbens are key in drug reward; salience network (dACC, insula) has been linked to craving and, notably, lesions in these regions have been associated with remission of addiction. - Pharmacological MRI and PET-fMRI work in animals suggests D1 receptor stimulation increases fMRI signals while D2 stimulation decreases them, supporting a rate-dependent receptor mechanism. - Clinical relevance: MP is widely prescribed for ADHD but has misuse potential, particularly via non-oral routes, underscoring the need to understand pharmacokinetic influences on brain function and reward.

Design: Double-blind, randomized, counterbalanced, within-subject study with three separate sessions per participant: (A) oral placebo + IV placebo, (B) oral MP 60 mg + IV placebo, (C) oral placebo + IV MP 0.25 mg/kg. Twenty healthy adults (mean age 36.1±9.6 years; 9 females), stimulant-naïve, completed the study (trial NCT03326245; IRB-approved). Sessions were separated by 40±35 days. Participants and staff were blinded; order was blocked across every six participants. Procedure and timing: At time 0 min, participants received the oral agent (MP or placebo). At 30 min, [11C]raclopride was administered as a bolus and simultaneous 90-min PET-fMRI acquisition began. At 60 min, participants received IV agent (MP or placebo) as a 30-s bolus. Throughout scanning, participants used a button box to rate 'How high do you feel?' (1–10). Ratings were sampled every 5 min from oral dosing; after IV dosing, every 1 min for 20 min, then every 5 min until end. Imaging acquisition: Simultaneous PET/MRI on a Siemens 3T Biograph mMR. PET: [11C]raclopride, 15.7±1.9 mCi bolus; list-mode for 90 min; OSEM reconstruction (3 iterations, 21 subsets), PSF modeling; 48 time frames (30×1 min, 12×2.5 min, 6×5 min). Attenuation correction via UTE-derived µ-maps with CNN-based method. SUV images normalized to body weight and dose; spatial normalization via HCP pipelines; SUVr(t) computed by normalization to cerebellar SUV. MRI: T1-weighted UTE for attenuation correction and MPRAGE for anatomy; EPI BOLD fMRI TE/TR=30/3000 ms; 3×3×4 mm voxels; 90 min acquisition. Preprocessing: HCP minimal pipelines and FSL for motion correction, field map processing, co-registration, and MNI normalization. Nuisance regression (white matter, CSF, global signals), 5 mm FWHM smoothing. For dynamic connectivity, bandpass 0.01–0.1 Hz. Behavioral and cardiovascular analyses: Repeated-measures ANOVA tested main effects of condition (oral MP, IV MP, placebo) and condition×time interactions for heart rate, systolic blood pressure, and 'high' ratings. PET analyses: - Static: Logan plot with cerebellum reference (t*=20 min) to derive DVR and BP_ND for putamen, caudate, ventral striatum. - Dynamic dopamine increases: Computed ASUVr(t) = SUVr_MP(t) − SUVr_placebo(t) and modeled with a gamma cumulative distribution function F(t) across subjects; its derivative f(t)=dF/dt estimated the rate of dopamine increases at 1-min resolution for oral and IV conditions. This approach leverages paired placebo scans to improve reliability and approximates LSSRM dynamics without requiring individual kinetic parameters. fMRI activity analysis: Whole-brain voxelwise multiple regression in SPM using PET-derived f(t) regressors (oral and IV; upsampled to fMRI TR via interpolation). A linear drift term was included. For each session, voxelwise maps of BOLD activity association with f(t) were generated. Group analyses: one-sample tests per condition and paired t-tests comparing oral vs placebo (slow dopamine increases) and IV vs placebo (fast dopamine increases). Drug-order covariate included. Control analyses added amplitude regressors to confirm rate-specific effects. Threshold: voxelwise p<0.001 uncorrected, cluster-level FDR p<0.05, k>50. Dynamic functional connectivity: Seeds were significant clusters from activation analyses to slow (oral>placebo) and fast (IV>placebo) increases. Sliding-window connectivity (5-min windows with 4-min overlap; producing 82 time points) computed as Fisher z-transformed correlations between seed BOLD and all voxels. Voxelwise regressions used oral and IV f(t) as regressors; group analyses as above. Primary seed highlighted: dACC (and left insula) from the fast-IV activation map. Sample size and power: Based on prior pharmacological imaging effect sizes, power analysis (G*Power 3.1.9.4) indicated n=20 to detect D=0.65 with α=0.05, power=0.8. Exploratory sex analyses were conducted but underpowered.

- Cardiovascular responses: Significant condition×time interactions for heart rate and systolic blood pressure, with fastest and largest increases after IV MP and modest, gradual changes after oral MP. Heart rate: F(80,2318)=3.022, p<2×10^-16; systolic BP: F(80,2277)=3.403, p<2×10^-16. Overall magnitude differed by condition for systolic BP (F(2,2277)=5.122, p=0.011) but not heart rate (F(2,2318)=2.746, p=0.077). - Static PET: BP_ND decreased relative to placebo for both oral and IV MP across striatal regions (corrected p<0.05), consistent with increased synaptic dopamine. No significant difference in BP_ND magnitude between oral and IV MP (F(1,85)=0.6; p=0.44), consistent with matched transporter occupancy (~70%). - Dynamic PET-derived dopamine kinetics: Oral MP produced earlier, slower, and smaller dopamine increases; IV MP produced later but faster and larger increases. The rate of dopamine increase f(t) (derivative of gamma CDF fit to ASUVr) captured these dynamics. - fMRI activity associations: - vmPFC showed significant decreases in BOLD activity associated with both slow (oral) and fast (IV) dopamine increases; its time course tracked the rate of dopamine increase across both routes. - Salience network regions (dACC and bilateral insula) showed significant BOLD increases selectively associated with fast (IV) dopamine increases and not with slow (oral) increases. - Dynamic functional connectivity: - Using the dACC cluster (fast-IV activation) as a seed, voxelwise analysis identified the dorsal caudate as the only region showing significant positive dynamic connectivity with dACC in association with the rate of fast dopamine increases. The left insula seed showed a similar dorsal caudate pattern. - The dACC–dorsal caudate connectivity time course was elevated selectively in the IV session and closely paralleled the PET-derived rate of dopamine increases. - Subjective 'high': Participants reported robust 'high' after IV MP; fewer reported 'high' after oral MP (only 13/20). The association between dACC–dorsal caudate connectivity and 'high' ratings was significant in the IV session (n=19 reporting 'high'; β=3.115, p=0.006), indicating that connectivity dynamics in this dorsal corticostriatal circuit track subjective reward. - Exploratory sex analyses: No significant sex differences were detected in maximal 'high', time-to-peak dopamine increases, baseline BP_ND, or strength of PET–fMRI associations (all p>0.22).

The findings demonstrate that the rate of dopamine increase, not merely its magnitude, determines which human brain circuits engage during stimulant exposure and how these relate to subjective reward. Fast dopamine increases induced by IV methylphenidate preferentially activated salience network nodes (dACC and insula) and strengthened their dynamic connectivity with dorsal caudate, a dorsal striatal target receiving dense dopaminergic input. This dorsal corticostriatal connectivity time course closely paralleled both PET-derived dopamine rate and subjective 'high', implicating this circuit in the conscious experience of drug reward when delivery is rapid. In contrast, vmPFC, a mesolimbic reward node, showed deactivation that tracked both slow and fast dopamine pharmacokinetics but did not correlate with 'high' in this stimulant-naïve sample, consistent with prior evidence that vmPFC responsivity may depend on expectation or sensitization after repeated drug exposure. The opposing BOLD responses across networks align with a receptor-based model: slow increases primarily engage inhibitory D2-mediated effects (decreases in activity), whereas fast increases additionally engage excitatory D1-mediated effects (increases in activity). The absence of significant striatal BOLD coupling to dopamine dynamics may reflect that hemodynamic effects predominantly manifest in projection targets (cortical-thalamic loops) rather than at striatal cell bodies, and that MP also modulates norepinephrine, further implicating frontal regions such as dACC that integrate convergent dopaminergic and noradrenergic inputs. The convergence with lesion network mapping—where remission of addiction aligns with downregulating dACC/insula-linked circuits and upregulating vmPFC-linked circuits—suggests therapeutic avenues targeting these networks. Overall, the study delineates a human dorsal corticostriatal circuit selectively sensitive to fast dopamine dynamics that underlie drug reward.

This work identifies a dorsal corticostriatal circuit—dACC/insula and their connectivity with dorsal caudate—that is selectively engaged by fast, but not slow, dopamine increases and closely tracks subjective 'high' after IV methylphenidate. vmPFC tracks dopamine pharmacokinetics across routes but does not predict subjective reward in stimulant-naïve individuals. By integrating simultaneous PET-fMRI with computational modeling of dopamine dynamics, the study reveals rate-dependent engagement of salience network circuitry in drug reward, providing mechanistic insight into why faster delivery routes have higher addiction potential. Future work should test causality by neuromodulating dACC–caudate circuitry during drug exposure to determine whether it attenuates subjective reward, assess generalization to substance use disorders and repeated drug exposure, and examine contextual and expectancy effects that may modulate vmPFC involvement.

- Participants were stimulant-naïve healthy adults; generalizability to individuals with substance use disorders and to repeated drug exposure is unknown. - Drug administration occurred in a tightly controlled laboratory environment, which has low ecological validity; contextual factors known to influence drug responses were not manipulated. - Sample size (n=20) limited power for subgroup analyses (e.g., sex effects) and detection of smaller effects. - Behavioral assessment focused on subjective 'high' as a proxy for reward; broader reward constructs (e.g., liking vs wanting, conditioning, alternative reinforcers) were not comprehensively measured. - Simultaneous PET-fMRI constraints precluded continuous radiotracer infusion; dynamic modeling relied on ASUVr with placebo scans and population-average fits, which may limit individual-level precision. - Striatal BOLD did not significantly couple with dopamine dynamics; interpretation relies on circuit-level projections and concurrent noradrenergic effects, which were not directly measured. - Expectancy effects were not manipulated and could differentially impact vmPFC responsivity.