Time-varying SUVr reflects the dynamics of dopamine increases during methylphenidate challenges in humans

Abstract

Dopamine facilitates cognition and is implicated in reward processing. Methylphenidate, a dopamine transporter blocker widely used to treat attention-deficit/hyperactivity disorder, can have rewarding and addictive effects if injected. Since methylphenidate's brain uptake is much faster after intravenous than oral intake, we hypothesize that the speed of dopamine increases in the striatum in addition to its amplitude underly drug reward. To test this we use simulations and PET data of [¹¹C]raclopride's binding displacement with oral and intravenous methylphenidate challenges in 20 healthy controls. Simulations suggest that the time-varying difference in standardized uptake value ratios for [¹¹C]raclopride between placebo and methylphenidate conditions is a proxy for the time-varying dopamine increases induced by methylphenidate. Here we show that the dopamine increase induced by intravenous methylphenidate (0.25 mg/kg) in the striatum is significantly faster than that by oral methylphenidate (60 mg), and its time-to-peak is strongly associated with the intensity of the self-report of "high". We show for the first time that the "high" is associated with the fast dopamine increases induced by methylphenidate.

Fig. 1. SUVr simulations: Specificity.

Time-varying fractional occupancy of dopamine transporter (DAT), $f_{occ}^{DAT}\left(t\right)$, and relative extracellular dopamine increases, D(t), were simulated using Eqns [6] and [8] and used as alternative mechanistic models for h(t), which is normalized to 1 and reflects the dynamics of binding competition between raclopride and dopamine increases induced by methylphenidate a. Noiseless time-varying differences in standardized uptake value ratios to the cerebellum (ΔSUVr) in striatum between placebo and intravenous methylphenidate (IV-MP) injected at t = 30 min, simulated using Eqns [2] and [3], showing the time delay (δ) between simulations with $h\left(t\right)\propto f_{occ}^{DAT}\left(t\right)$ or $D\left(t\right)$ b. The dynamics of the normalized $f_{occ}^{DAT}\left(t\right)$ did not resemble the dynamics of simulated ΔSUVr (dots) when $h\left(t\right)\propto f_{occ}^{DAT}\left(t\right)$ but that of D(t) did it when $h\left(t\right)\propto D\left(t\right)$ c, d. Normal random noise (3%) was added to the SUVr time courses for placebo and IV-MP prior to compute ΔSUVr in c and d. $K_1^{MP}$ = 0.6 min⁻¹, $k_2^{MP}$ = 0.06 min⁻¹, $k_3^{MP}$ = 0.5 min⁻¹, and $k_4^{MP}$ = 0.2 min⁻¹ (see ref. ); K_1r = 0.092 mL/min g, k_2r = 0.45 min⁻¹, and R₁ = 1.154, k₂ = 0.45 min⁻¹, and k_2a = 0.065 min⁻¹ (see ref. ); β = 0.02 min⁻¹.

Fig. 2. SUVr simulations: Sensitivity.

Dynamic dopamine increases, D(t), simulated with Eqns [6] and [8], and the corresponding changes in standardized uptake value ratio (ΔSUVr) between placebo and intravenous methylphenidate (IV-MP) simulated with Eqns [2] and [3] for h(t)=D(t) and 3 different pharmacological doses of MP. Normal random noise (3%) was added to the SUVr time courses for placebo and IV-MP prior to compute ΔSUVr. $K_1^{MP}$ = 0.6 min⁻¹, $k_2^{MP}$ = 0.06 min⁻¹, $k_3^{MP}$ = 0.5 min⁻¹, and $k_4^{MP}$ = 0.2 min⁻¹ (see ref. ); K_1r = 0.092 mL/min g, k_2r = 0.45 min⁻¹, and R₁ = 1.154, k₂ = 0.45 min⁻¹, and k_2a = 0.065 min⁻¹ (see ref. ); β = 0.02 min⁻¹.

Fig. 3. SUVr simulations: Curve fitting.

Extracellular dopamine increases, D(t), simulated with Eqns [6] and [8], the corresponding changes in standardized uptake value ratio (ΔSUVr) between placebo and intravenous methylphenidate, simulated with Eqns [2] and [3] for h(t)=D(t), and a curve fit to the ΔSUVr data using a gamma cummulative distribution (CDF), F(t), given by Eqn [9] a. For 0.25 mg/kg methylphenidate, the time derivatives of D(t) and F(t), d(t) and f(t), had similar time-to-peak (TTP) b. Normal random noise (3%) was added to the SUVr time courses for placebo and IV-MP prior to compute ΔSUVr in a. $K_1^{MP}$ = 0.6 min⁻¹, $k_2^{MP}$ = 0.06 min⁻¹, $k_3^{MP}$ = 0.5 min⁻¹, and $k_4^{MP}$ = 0.2 min⁻¹ (see ref. ); K_1r = 0.092 mL/min g, k_2r = 0.45 min⁻¹, and R₁ = 1.154, k₂ = 0.45 min⁻¹, and k_2a = 0.065 min⁻¹ (see ref. ); β = 0.02 min⁻¹.

Fig. 4. SUVr simulations for intravenous MP: Variability and accuracy in TTP:.

Scatter plots showing the lack of significant associations between time-to-peak (TTP) of the fitted gamma probability density functions, f(t), and the parameters in Eqns [2], [3], and [6] (K₁, R₁, k₂, k_2a, A₁, A₂, A₃, T₁, T₂, and T₃; see Methods), which were randomly varied 4% within- and 10% between-simulations (N = 1000) with h(t)=D(t); differently, fitted TTP was sensitive to 10% random variations in the dose and the decay rate of the concentration of methylphenidate (MP) in blood, λ a. Fitted TTP was linearly associated across 1000 simulations with true TTP of the rate of D(t) b. The histogram shows the skewed distribution of the TTP difference between fitted and dopamine rate TTP, dTTP c. Normal random noise (3%) was added to the SUVr time courses for placebo and IV-MP before computing ΔSUVr. $K_1^{MP}$ = 0.6 min⁻¹, $k_2^{MP}$ = 0.06 min⁻¹, $k_3^{MP}$ = 0.5 min⁻¹, and $k_4^{MP}$ = 0.2 min⁻¹ (see ref. ); K_1r = 0.092 mL/min g, k_2r = 0.45 min⁻¹, and R₁ = 1.154, k₂ = 0.45 min⁻¹, and k_2a = 0.065 min⁻¹ (see ref. ); β = 0.02 min⁻¹.

Fig. 5. SUVr simulations for oral MP: Variability and accuracy in TTP.

Scatter plots showing the lack of significant associations between time-to-peak (TTP) of the fitted gamma probability density functions, f(t), and the parameters in Eqns [2] and [6] (K₁, R₁, k₂, k_2a;, and t_peak see Methods), which were randomly varied 4% within- and 10% between-simulations (N = 1000) with h(t)=D(t); differently, fitted TTP was sensitive to 10% random variations in t_peak and methylphenidate (MP) dose a. Fitted TTP was linearly associated across 1000 simulations with true TTP of the rate of D(t) b. The histogram shows the skewed distribution of the TTP difference between fitted and dopamine rate TTP, dTTP c. Normal random noise (3%) was added to the SUVr time courses for placebo and oral-MP before computing ΔSUVr. $K_1^{MP}$ = 0.6 min⁻¹, $k_2^{MP}$ = 0.06 min⁻¹, $k_3^{MP}$ = 0.5 min⁻¹, and $k_4^{MP}$ = 0.2 min⁻¹ (see ref. ); K_1r = 0.092 mL/min g, k_2r = 0.45 min⁻¹, and R₁ = 1.154, k₂ = 0.45 min⁻¹, and k_2a = 0.065 min⁻¹ (see ref. ); β=0.02 min⁻¹.

Fig. 6. SUVr dynamics in humans.

a Differences in static standardized uptake value ratio (ΔSUVr) in the putamen (relative to the cerebellum) as a function of differences in non-displaceable binding potential (BPnd) between placebo (PL) and methylphenidate (MP) scans, for intravenous (IV) and oral sessions, and statistical t-score maps reflecting differences in BPnd between placebo and MP conditions, superimposed on axial views of the human brain at the level of the striatum. b Average ΔSUVr time courses (dots), and fitted gamma cumulative distribution (F) and probability (f) functions across 20 healthy adults for intravenous (IV) and oral MP. The arrow highlights the time-to-peak (TTP) of f(t) since the onset of MP administration.

Fig. 7. Association between shorter TTP and stronger “high” ratings.

a Average high ratings across 20 participants as a function of time during the scans. b Shorter time-to-peak (TTP) was associated with stronger differences in peak “high” ratings between methylphenidate (MP) and placebo (PL), independently for oral (n = 13) and intravenous (IV; n = 18) MP.