European Journal of Nuclear Medicine and Molecular Imaging

, Volume 37, Issue 12, pp 2364–2370

Levodopa and pramipexole effects on presynaptic dopamine PET markers and estimated dopamine release


    • University of British Columbia
    • Department of Physics and Astronomy
  • Katherine Dinelle
    • University of British Columbia
  • Michael Schulzer
    • Department of Physics and Astronomy
  • Edwin Mak
    • Department of Physics and Astronomy
  • Doris J. Doudet
    • University of British Columbia
  • Raúl de la Fuente-Fernández
    • University of British Columbia
    • Department of Physics and Astronomy
Original Article

DOI: 10.1007/s00259-010-1581-3

Cite this article as:
Sossi, V., Dinelle, K., Schulzer, M. et al. Eur J Nucl Med Mol Imaging (2010) 37: 2364. doi:10.1007/s00259-010-1581-3



Levodopa and dopamine (DA) agonist therapy are two common treatments for Parkinson’s disease (PD). There is controversy about the effects of these treatments on disease progression and imaging markers. Here we used multi-tracer positron emission tomography imaging and a unilateral 6-hydroxydopamine (6-OHDA) rat model of PD to evaluate in vivo the effects of chronic levodopa and pramipexole treatments on measurements of vesicular monoamine transporter type 2 (VMAT2), dopamine transporter (DAT) levels, and on levodopa-induced changes in synaptic DA levels [Δ(DA)].


Twenty-three unilaterally 6-OHDA lesioned rats underwent an 11C-dihydrotetrabenazine (DTBZ, VMAT2 marker), an 11C-methylphenidate (MP, DAT marker), and a double 11C-raclopride (RAC, D2-type receptor marker) scan. They were assigned to three treatment groups: saline (N = 7), pramipexole (N = 8), and levodopa (N = 8). After 4 weeks of treatment, imaging was repeated.


Results showed (1) a significant treatment effect on DTBZ, with pramipexole decreasing DTBZ binding compared to levodopa, (2) significant side and treatment-striatal side interaction effects for MP, indicating that levodopa tends to decrease MP binding compared to pramipexole, and (3) no treatment effect on Δ(DA).


These data indicate that while chronic dopaminergic pharmacological treatment affects DTBZ and MP binding, it does not affect levodopa-induced changes in synaptic DA level.


DATDopamine agonistLevodopa6-OHDA rodent modelParkinson’s diseaseVMAT2


Parkinson’s disease (PD), which affects approximately 300/100,000 of the general population [1], is characterized by a progressive degeneration of the dopaminergic system leading to increasingly more severe dopamine (DA) deficiency. A thorough understanding of disease origin and of the neurochemical changes associated with disease initiation and progression is still elusive. No cure exists at present; therapeutic approaches are aimed at enhancing synaptic dopamine levels to mitigate the effects of impaired neurotransmission. Levodopa and DA agonists are currently the two most common treatment options. They both appear effective in the early stages, while generally leading to motor and psychiatric side effects as disease progresses. Currently there is no agreement on the effect of treatment on the progression of the dopaminergic deficit nor is there consensus on how each treatment might influence the probability of side effects.

Several clinical and imaging studies have been designed to test efficacy of different treatments with varying outcomes. A common difficulty in such studies is the identification of objective markers of disease progression. Imaging is often used to that purpose. Yet, results from imaging studies are not always consistent with clinical observations. For example, a multicenter study performed in a large group of PD patients compared the efficacy of a DA agonist to that of levodopa therapy using single photon emission computed tomography (SPECT) DA transporter (DAT) imaging and clinical outcomes [2, 3]. SPECT data showed a relative preservation of DAT binding in patients on agonist therapy, which was interpreted as slowing of disease progression, which was found not to be consistent with clinical measures. This study was subsequently criticized with the observation that results obtained from DAT imaging might not correctly represent disease severity, owing to the high susceptibility of DAT to pharmacological regulation [4, 5]. A similar rationale was also presented in an earlier human study [6], where short-term levodopa treatment was found to be associated with downregulation of striatal DAT in early PD patients.

Concern about a direct treatment impact on the imaging markers themselves is not limited to DAT imaging. 11C-dihydrotetrabenazine (DTBZ) binding has been until very recently considered a more accurate marker of disease severity; animal studies have shown excellent correlations with striatal tyrosine hydroxylase (TH) levels and no susceptibility to pharmacological regulation [79]. The latter view has been recently contradicted by findings that demonstrated that DTBZ binding is affected by pharmacological manipulation of vesicular DA concentration [1012]. These findings highlight the importance of investigating the pharmacological effects of treatment on the presynaptic imaging markers before they can be used as surrogate measurements of disease progression.

Another controversial aspect is the influence of each treatment type on the propensity of developing motor complications. Current understanding claims that motor complications in the form of wearing off phenomena and dyskinesias are associated with a combination of dopaminergic neuronal loss and altered DA release patterns, which lead to large oscillation in synaptic DA levels and thus pulsatile stimulation of postsynaptic receptors[1315]. This, in turn, may lead to downstream changes in proteins and gene expression. It is thus plausible to speculate that a mechanism, by which a treatment could influence the probability of developing motor complications, is by altering DA release patterns, i.e., inducing higher treatment-related DA release. Considering that patients on levodopa therapy seem to suffer from a higher incidence of motor complications [16] there is belief, although controversial [13, 17], that levodopa therapy itself could induce alterations in the DA release pattern, increasing the predisposition to motor complications compared to DA agonist therapy.

This study was thus designed to investigate two specific questions. First we studied the impact of treatment on vesicular monoamine transporter type 2 (VMAT2) and DAT binding as estimated by DTBZ and 11C-methylphenidate (MP) imaging, and secondly we investigated if either treatment influences DA release patterns after levodopa challenge.

In order to address these questions, we used a unilateral 6-hydroxydopamine (6-OHDA) rat model of PD where the unlesioned side was used as control. While several similar studies have already been done with this animal model of PD also leading to inconsistent results, the unique aspect of the present study was the use of multi-tracer longitudinal PET imaging with protocols similar to those used in human imaging. The studies were initiated at least 4 weeks after lesioning to minimize effects of possible acute lesion-induced adaptive changes; furthermore a group of saline-treated rats was used as a control to account for possible residual lesion-induced effects. DA release patterns were investigated by assessing levodopa-derived changes in synaptic DA levels using a double 11C-raclopride (RAC) scanning protocol (see “Materials and methods” section), a well-validated approach in both human and rodents. While we previously reported pre-treatment imaging results for some of these animals [18], here we focus on the effect of chronic treatment on the imaging markers.

Materials and methods

Subjects, animal lesioning, and treatment

Twenty-three 6-OHDA unilaterally lesioned male 400–450 g Sprague-Dawley rats were included in this study. The animals were anesthetized with isoflurane and placed in the stereotaxic head holder with the skull flat between lambda and bregma. A solution of 6-OHDA hydrobromide (8 μg in 4 μl 0.05% ascorbic acid in saline; Sigma) was infused at two sites along the medial forebrain bundle [site 1: AP -2.8 mm, ML -1.8 mm, DV -8.0 (all from bregma); site 2: AP -4.7 mm (from bregma), ML -1.5 mm (from midline), DV -7.9 mm (from skull)] according to Paxinos and Watson [19].

Approximately 30 min prior to the 6-OHDA infusion the subjects were given desipramine (25 mg/kg i.p.) to protect noradrenergic fibers [2022]. The 6-OHDA solution was delivered at a rate of 1 μl/min using a 50 μl Hamilton syringe through a 26-gauge needle. Following the injection of 6-OHDA, the needle was left in place for an additional 3 min prior to withdrawal. After surgery, rats were monitored and then returned to their home cage only after full recovery from the anesthesia. The animals were allowed 4 weeks recovery before undergoing the PET studies.

After the completion of the pre-treatment set of scans (an MP, a DTBZ, and two RAC scans), rats were assigned into three treatment groups matched for degree of dopaminergic denervation (DD) as assessed from the DTBZ scan (see below): pramipexole (N = 8, DD = 0.74 ± 0.18, range: 0.50–0.96), levodopa (N = 8, DD = 0.72 ± 0.18, range: 0.50–0.96), and saline (N = 7, DD = 0.79 ± 0.14, range: 0.54–0.96). All rats were injected i.p. once a day with 10 mg/kg of levodopa mixed with 15 mg/kg benserazide (levodopa group) or 1 mg/kg per day of pramipexole (pramipexole group) or saline (saline group) for a minimum of 4 weeks (35 ± 13 days, mean ± SD). Four weeks was considered long enough for the system to have recovered from any acute treatment-induced changes and to have stabilized; the treatment doses are commonly used in rat studies. In all cases, the animals were scanned 24 h after the last treatment administration to maximize the sensitivity to chronic treatment-related effects. While the half-life of pramipexole in rats is not exactly known, 24 h were deemed sufficient to minimize any acute effects of pramipexole [23].

PET scanning

Each rat underwent four scans, at baseline (pre-treatment) and after chronic treatment: one DTBZ, one MP, and two RAC scans, the first (RAC0) before and the second 45 min after i.p. administration of levodopa/benserazide (50 mg/kg mixed with 15 mg/kg) (RAC1).

The scanning procedure was identical for all three tracers in all conditions. The rats were anesthetized with a 2% isoflurane gas mixture. All studies were performed on a Siemens/Concorde Focus 120 with a resolution of (1.7 mm)3 [24]. Following a 10-min transmission scan with a 57Co source, a 1-h emission scan was performed starting at tracer injection. Data were histogrammed into 6 × 30 s, 2 × 60 s, 5 × 300 s, 2 × 450 s, and 2 × 480 s frames and reconstructed using Fourier rebinning and filtered backprojection after applying normalization, scatter, attenuation, and sensitivity corrections. The specific activity (SA) for each tracer was kept high enough to ensure negligible transporter or receptor occupancy (SA > 1,100 nCi/pmol) at an injection dose of 100 μCi/100 g of animal weight [25, 26].

Image and data analysis

Identical analysis methods were used for the pre- and post-treatment scans following a method described in detail previously [25]. Briefly, regions of interest (ROI) were placed on both striata and the cerebellum with the aid of a coregistered rat brain atlas. An ROI (area 9.00 mm2) was placed on each striatum on three consecutive axial planes (for a total of three ROIs per striatum) and a reference ROI (area 13.50 mm2) was placed on the cerebellum on three consecutive axial planes. Time-activity curves (TAC) were generated for each ROI and averaged across all three planes. Tissue input binding potential BPND values were calculated from the average TACs by applying the simplified reference tissue method [27, 28] to each striatum separately.

Dopaminergic denervation

Data from the pre-treatment DTBZ scan were used to determine the level of DD. A good correlation has previously been found between DTBZ binding and TH activity in the striatum, indicating that DTBZ binding is a good measure of striatal dopamine terminal density [7]. DD was estimated as
$$ {\text{DD = }}{{\left( {{\text{BP}}_{{{\text{ND\_DTBZ\_C}}}} {\text{ - BP}}_{{{\text{ND\_DTBZ\_L}}}} } \right)}} \mathord{\left/ {\vphantom {{{\left( {{\text{BP}}_{{{\text{ND\_DTBZ\_C}}}} {\text{ - BP}}_{{{\text{ND\_DTBZ\_L}}}} } \right)}} {{\text{BP}}_{{{\text{ND\_DTBZ\_C}}}} {\text{ $ \times $ 100}}}}} \right. \kern-\nulldelimiterspace} {{\text{BP}}_{{{\text{ND\_DTBZ\_C}}}} {\text{ $ \times $ 100}}} $$
where L and C refer to the lesioned and control side, respectively. A difference in DTBZ BPND between the control and lesioned side was previously demonstrated to reflect a change in the maximum free VMAT2 density Bmax and not ligand-transporter affinity KD, justifying this definition of DD [25]. We have also shown an excellent correlation between the DD levels assessed with PET and those assessed with autoradiography using the same tracer [18], thus demonstrating that possible partial volume effects do not adversely impact the PET measurement.

The highest level of DD estimated by PET in this rat population was 96%, with a corresponding value of 98% estimated from autoradiography [18]; this indicates that any potential confound arising from DTBZ binding to serotonergic terminals [7], which are unaffected by 6-OHDA lesioning [29], must be negligible.

Change in synaptic DA levels [Δ(DA)]

An estimate of the levodopa-derived changes in synaptic DA levels was obtained by comparing the BPND values obtained in the first (baseline) RAC scan (RAC_0) (BPND_RAC_0) and those obtained in the second RAC scan (RAC_1) performed after challenge administration of levodopa/benserazide (BPND_RAC_1). Such an estimate can be performed according to the accepted definition [30]:
$$ \Delta {\left( {{\text{DA}}} \right)}{\text{ = }}{{\left( {{\text{BP}}_{{{\text{ND\_RAC\_0}}}} {\text{ - BP}}_{{{\text{ND\_RAC\_1}}}} } \right)}} \mathord{\left/ {\vphantom {{{\left( {{\text{BP}}_{{{\text{ND\_RAC\_0}}}} {\text{ - BP}}_{{{\text{ND\_RAC\_1}}}} } \right)}} {{\text{BP}}_{{{\text{ND\_RAC\_0}}}} {\text{ $ \times $ 100}}}}} \right. \kern-\nulldelimiterspace} {{\text{BP}}_{{{\text{ND\_RAC\_0}}}} {\text{ $ \times $ 100}}} $$

Statistical analysis

Care was taken to isolate the treatment-related effect from a possible lingering lesion-induced effect. This was accomplished by first linearly regressing the changes from baseline in the saline group on the corresponding baseline saline values for each side and marker separately, and then, using the resulting prediction equation [predicted Δ(marker)], adjusting changes from baseline in the treatment groups for their saline-expected relation [Δ(marker)] as follows:
$$ \Delta ({\hbox{marker}}){ } = ( {\hbox{ marke}}{{\hbox{r}}_{\rm{Post}}}--{\hbox{ marke}}{{\hbox{r}}_{\rm{Pre}}} ) --{\hbox{ predicted}}\Delta ({\hbox{marker}}) $$

For each marker, an analysis of covariance (ANCOVA) was used to assess the relation between the resulting saline-adjusted values Δ(marker) and (1) their corresponding observed baseline values, (2) side (control or lesion), (3) treatment (pramipexole or levodopa), and (4) the respective, second-order interaction terms. F tests were used to compare the treatment effects as part of the covariance model that we employed. The contribution of each explanatory variable was considered significant whenever p ≤ 0.05.


Changes in BPND_DTBZ

Pre-treatment BPND_DTBZ values are listed in Table 1. There was a significant treatment effect (p = 0.05), with the pramipexole group showing a greater BPND_DTBZ decrease compared to the levodopa group (Fig. 1, Table 2). No effect of striatal side was found for DTBZ binding. No significant correlation between the Δ(BPND_DTBZ) and pre-treatment BPND_DTBZ was found. Regression results for the saline group were: −0.50 (slope), 1.71 (intercept), and r = 0.55 for the control side and 0.16 (slope), −0.075 (intercept), and r = 0.72 for the lesioned side.
Table 1

Initial, pre-treatment value averages across the three groups. (There was no statistical difference between the three groups)


Lesioned side

Control side

BPND_DTBZ (mean ± SE)

0.95 ± 0.13

3.81 ± 0.08

BPND_MP (mean ± SE)

0.32 ± 0.03

1.16 ± 0.03

Δ(DA) (%) (mean ± SE)

10 ± 3

1.2 ± 1.6
Fig. 1

Changes in BPND_DTBZ [BPND_DTBZ (post-treatment) − BPND_DTBZ (pre-treatment)] for the three treatment groups, pramipexole (PRA), levodopa (LD), and saline (SAL). Bars represent standard errors. In the figure, changes are shown for each treatment group separately and are not adjusted for the predicted values (see text)

Table 2

Δ(marker), i.e., changes in the parameter values in two treatment groups adjusted for the predicted change as determined from the saline group (see text). Data show mean ± SE. The average adjusted change in the saline group is zero, by definition. Since only the treatment effect was significant for Δ(BPND_DTBZ), average values over both sides are presented





Lesioned side

Control side


−0.11 ± 0.06

−0.01 ± 0.03

−0.02 ± 0.02


0.07 ± 0.06

0.02 ± 0.03

−0.08 ± 0.05

PRA pramipexole, LD levodopa

Changes in BPND_MP

Pre-treatment values are listed in Table 1. A significant striatal side (p = 0.003) and treatment-striatal side interaction effect (p = 0.04) was observed for MP binding, with the levodopa-treated group showing a larger decrease on the control side as compared to the lesioned side than the pramipexole-treated group (Fig. 2, Table 2). As shown in Fig. 2, a clear trend towards an MP binding increase was observed on the lesioned side (p = 0.06 when all three treatment groups were taken together). A significant positive correlation between the Δ(BPND_MP ) and pre-treatment (baseline) BPND_MP was also found (slope = 0.32; p = 0.003). Regression results for the saline group were: −0.33 (slope), 0.39 (intercept), and r = 0.45 for the control side and −0.30 (slope), 0.11 (intercept), and r = 0.78 for the lesioned side.
Fig. 2

Changes in BPND_MP [BPND_MP (post-treatment) − BPND_MP (pre-treatment)] for the three treatment groups, pramipexole (PRA), levodopa (LD), and saline (SAL) for the lesioned and control sides. Bars represent standard errors. In the figure, changes are shown for each treatment group separately and are not adjusted for the predicted values (see text)

Changes in dopamine release [Δ(DA)]

No significant effect of any kind was observed for Δ(DA) (Table 1) with final, post-treatment values averaged across the three groups being 9 %± 3% (mean ± SE) on the lesioned side and 0.9 %± 1.4% (mean ± SE) on the control side.


Our study provides the following findings (1) DTBZ binding is affected differently by pramipexole and levodopa, independently of the level of dopaminergic denervation, indicating that VMAT2 binding is susceptible to pharmacological regulation; (2) MP binding is decreased by levodopa in a denervation severity-dependent fashion, with a greater reduction of DAT binding observed for a more intact system; and (3) RAC changes indicate that levodopa-induced dopamine release patterns are not affected by chronic treatment.

DTBZ binding

Two observations can be highlighted. No effect of denervation severity was observed on the post- to pre-treatment BPND_DTBZ comparison, which supports the notion that DTBZ binding and thus VMAT2 is likely not subject to long-term lesion (disease)-induced regulation [31]. A treatment effect was however observed with pramipexole decreasing DTBZ binding more that levodopa. VMAT2 binding seems therefore to be susceptible to pharmacological regulation. The most likely mechanism for this finding is that pramipexole increases vesicular DA levels [32]. Previous studies have indeed found that DTBZ binding is sensitive to the vesicular DA concentration levels [1012]. Observed decreases in DTBZ binding were approximately 8%, which is consistent with the reported studies. Additionally, pramipexole was shown to decrease cell firing [23, 33], further contributing to increased vesicular DA levels.

MP binding

Several observations can be made here. Changes in MP binding were found to be related to striatal side, pre-treatment binding values, and a treatment-striatal side interaction, with levodopa treatment inducing a greater decrease in DAT binding on the control side compared to pramipexole. Our results also indicate that there is a denervation-dependent increase in DAT binding on the lesioned side. The latter observation probably reflects a compensatory upregulation of DAT expression, aimed at preserving DA levels in a situation of DA deficiency, which is consistent with recent studies [34, 35]. In other words, increased DAT expression would be aimed at preventing an excessive extrasynaptic diffusion and metabolism of DA upon cell firing and thus extend the half-life of striatal DA [36]. Compensatory DAT expression upregulation on the lesioned side is probably mediated by Nurr1; it is known that Nurr1 activation increases the activity of TH, the rate-limiting enzyme for endogenous dopamine synthesis, and also induces DAT expression [37].

Taken together, these findings indicate that levodopa by itself (i.e., on the control side, where the denervation-induced effect is not dominant) tends to decrease MP binding more than pramipexole at these dose levels. This treatment effect can be explained by a differential effect of levodopa and pramipexole on dopamine D2/D3 autoreceptors [38]. Animal experiments have shown that levodopa induces dopamine D3 receptor upregulation [39], and it is now known that dopamine D3 receptors are in fact autoreceptors [40]. Pramipexole, on the other hand, has exactly the opposite effect, D2/D3 autoreceptor desensitization [23]. As the activation of dopamine D2/D3 autoreceptors leads to a decrease in both dopamine release and dopamine reuptake [36, 41, 42], DAT expression is expected to be lower in levodopa-treated animals compared to the pramipexole group. In addition, a levodopa-induced relative excess of DA on the control side could potentially induce a downregulation of DAT expression through Nurr1 deactivation [37].

Although these conclusions cannot be directly extrapolated to human disease, since the 6-OHDA PD model is relatively static and not progressive, our results seem consistent with previously reported human studies [2, 6], where the effects of levodopa and pramipexole on DAT were assessed in early stages of the disease and thus relatively preserved dopaminergic function.

Changes in levodopa-derived synaptic DA levels

No change was observed in DA release after levodopa challenge in any of the three groups. This finding shows that chronic levodopa treatment does not alter DA release patterns as sampled with this imaging protocol. To the extent to which altered DA release is involved in the genesis of motor complications, levodopa treatment by itself does not seem to enhance the probability of developing motor complications. These results are in keeping with human studies showing that dopaminergic treatment does not contribute to treatment-related pharmacological priming [13] and also with recent findings obtained using chronoamperometry techniques in rodents [43]. Although clinical observations show that PD patients on levodopa therapy have a higher incidence of dyskinesia than patients on dopamine agonists, the key point is that they are more likely to experience dyskinesia rather than being more prone to developing dyskinesias as disease progresses.

In terms of imaging markers, our results indicate that VMAT2 binding is an objective and accurate marker of lesion (disease) severity in the absence of pharmacological intervention. This contrasts with DAT binding, which is susceptible to both lesion-related and pharmacologically related regulation. Nonetheless, it should be noted that even DTBZ binding might not correctly quantify treatment effects.

Our results caution against the assumption that the effect of a pharmacological treatment on the lesioned side can be evaluated by direct normalization to the control side, as is often done in postmortem animal studies, where longitudinal observations are not possible. Thus, for example, such an approach applied to the present data can incorrectly lead to the conclusion that levodopa has a neuroprotective effect, significantly reducing the level of denervation severity as estimated by MP binding [44]. However, in reality, such a change in MP binding only reflects a striatal side effect (Fig. 2), not related to neuroprotection.

Our findings indicate that treatment efficacy studies need to be designed with a great amount of caution, and bearing in mind potential confounds arising from the interaction between treatment and selected observables.


This work was supported by CIHR, NSERC, MSFHR and Triumf Life Science.

Conflicts of interest


Copyright information

© Springer-Verlag 2010