In vivo mesolimbic D2/3 receptor binding predicts posttherapeutic clinical responses in restless legs syndrome: a positron emission tomography study

Abstract

Although D2/3 agonists have been used as a first-line medication for idiopathic restless legs syndrome (iRLS), findings on D2/3 receptors have been inconsistent. Here, we aimed to clarify the contribution of D2/3 receptor function to the clinical symptoms of iRLS by comparing the binding potential (BP_ND) of [¹¹C]raclopride with clinical improvements after D2/3 stimulation by pramipexole. Eight drug-naïve, iRLS patients and eight age-matched healthy subjects were scanned with positron emission tomography (PET). After PET scans, all patients received pramipexole (0.125 mg) orally for 2 weeks. Patients were evaluated every day with several standardized clinical tests. The BP_ND values were compared using regions of interest and voxel-based methods. Results showed that the mean magnitude of [¹¹C]raclopride BP_ND in the mesolimbic dopamine region (nucleus accumbens (NA) and caudate) was significantly lower in the iRLS group. No significant differences between groups were observed in the putamen. The NA [¹¹C]raclopride BP_ND levels correlated negatively with clinical severity scores and positively with the degree of posttreatment improvement in iRLS. The present results suggest that alterations in mesolimbic D2/3 receptor function reflect the pathophysiology of iRLS, and the baseline availability of D2/3 receptors may predict the clinical outcome after D2/3 agonist treatment.

Introduction

Idiopathic restless legs syndrome (iRLS), characterized by the irresistible, often indescribably unpleasant urge to move the limbs at rest, has a more subtle onset than that of the secondary form of RLS (Hanson et al, 2004). As such, it is often misdiagnosed, leading to a postponement in proper medication. Many iRLS cases are believed to be hereditary, but there are also a number of sporadic cases (Hening et al, 1999). The diagnosis of iRLS is based on clinical symptoms and serological tests, excluding factors relevant to secondary RLS such as a deficiency in ferritin or iron binding, and may be confirmed when a patient responds well to dopaminergic medication. This type of diagnostic trial approach is reminiscent of conventional clinical procedures for the diagnosis of Parkinson's disease. Indeed, RLS patients, irrespective of idiopathic or secondary types, mostly benefit from dopaminergic therapy (Allen, 2004). However, unlike Parkinson's disease, dopaminergic cell loss is not found in iRLS (Pittock et al, 2004) and a reduction in dopamine content has not been observed in the postmortem brains of iRLS patients (Satija and Ondo, 2008).

Although D2/3 agonists are currently regarded as first-line medications for iRLS, there have been inconsistent reports on D2/3 function. It remains unclear as to how much and where in the dopamine system the D2/3 receptor contributes to the amelioration of iRLS symptoms after treatment. Previous in vivo studies with tracers for dopamine synthesis and dopamine transporter ligands reported that presynaptic dopamine function is not significantly altered in RLS (Ruottinen et al, 2000; Trenkwalder et al, 1999; Turjanski et al, 1999). In contrast, D2/3 receptor imaging using [¹¹C]raclopride and [¹²⁵I]iodobenzamide yields conflicting results: decreases in tracer binding (Michaud et al, 2002; Turjanski et al, 1999), no changes in binding (Eisensehr et al, 2001; Tribl et al, 2004), or increases in binding (Cervenka et al, 2006) in different RLS patient populations. However, these studies were conducted in mixed populations, consisting of patients with or without a family history, and also included patients with a history of pharmacological treatment, which further complicates the interpretation of results. The presence of structural and functional changes in dopamine receptors is dependent on genes that regulate the expression of receptors, and also upon the history of exposure to medication that potentially alters receptor availability (Missale et al, 1998). Therefore, in studies focusing on the innate pathophysiological features of iRLS assessed by molecular imaging methods, it is desirable to exclude patients taking medication.

The purpose of the present study was to clarify the relationship between D2/3 receptor binding and D2/3 agonist-induced clinical responses in drug-naive iRLS patients. To achieve this, the binding potential (BP_ND) of [¹¹C]raclopride, a marker for the D2/3 receptor, was measured in different areas of the mesolimbic (nucleus accumbens (NA) and caudate) and nigrostriatal (ventral and dorsal putamen) dopaminergic projection regions using positron emission tomography (PET), and binding levels were compared with clinical responses after D2/3 agonist stimulation by pramipexole.

Subjects and methods

Participants

We recruited eight drug-naive, nonhereditary iRLS patients (three males, five females, 58.4 years±12.0 s.d.) diagnosed with the International Restless Legs Syndrome Study (IRLSS) Group diagnostic criteria and exclusion criteria, from which patients who had a family history, and/or previous medication history of the use of dopaminergic agents, benzodiazepines, opioids, anticonvulsants, and/or were suffering from neurological or psychiatric disorders were excluded. Age-matched healthy controls (three males, five females, 56.0 years±14.8 s.d.) were also examined. Subjects provided written informed consent to participate in the present study, which was approved by the local ethics committee of Hamamatsu Medical Center. All participants underwent magnetic resonance imaging, a neuropsychiatric examination, and a blood test to exclude the possibility of any accompanying disease. In particular, ferritin and hematocrit levels had to be within normal limits in all patients.

Clinical Assessments and Drug Intervention

The clinical severities of iRLS patients were assessed with the IRLSS Group rating scale for RLS symptoms, the Japanese version of the ESS (Epworth Sleepiness Scale) (Takegami et al, 2009) for daytime sleepiness, and the Pittsburgh Sleep Quality Index (PSQI) (Doi et al, 2000) for subjective sleep quality. In the current study, the IRLSS scores varied from 9 to 36, indicating that RLS symptoms ranged from mild to severe. After completion of the PET examination, patients started treatment with the dopamine D2/3 agonist pramipexole at a (nightly) dose of 0.125 mg per day for 1 week, followed by up to 0.25 mg/day for another week in some patients, during which time patients were asked to complete the above-mentioned clinical assessments every day. We used data of the IRLSS taken at entry and at day 14 to calculate the degree of the therapeutic response by pramipexole and to examine the relationship between the responses and levels of receptor availability evaluated by PET. The clinical response was calculated as follows: Response (%)=−(score^T=14−score^T=0)/score^T=0 × 100, where IRLS scores were taken at entry (T=0) and at 14 days after pramipexole treatment (T=14). Clinical characteristics of patients are shown in Table 1.

Table 1. Characteristics of idiopathic restless legs syndrome patients.

No.	Age (years)	Sex	DD (years)a	Past history	Medication	Clinical scores			Responseb
						IRLSS	ESS	PSQI	IRLSS
1	57	F	11	Lung cancerc	(−)	28	8	2	14.3
2	31	F	0.8	(−)	Vitamins	36	8	12	22.2
3	62	M	0.3	(−)	(−)	26	5	17	15.4
4	64	F	1	Cholelithiasis	(−)	39	1	3	15.4
5	66	M	0.33	Limb fracture	(−)	13	12	13	30.8
6	69	F	10	(−)	(−)	9	3	17	22.2
7	58	F	0.25	(−)	(−)	27	11	8	25.9
8	63	M	1.1	(−)	(−)	25	9	14	20.0

IRLSS, International Restless Legs Syndrome Study Group rating scale, Epworth sleepiness scale (ESS), Pittsburgh Sleep Quality Index (PSQI) measured at entry; PET, positron emission tomography; DD, disease duration.

^aThe disease duration (years) between disease onset and PET examination.

^bResponse (%)=−(score^T=14−score^T=0)/score^T=0 × 100.

^cA well-differentiated lung cancer was totally resected over 15 years ago.

Magnetic Resonance Imaging and Positron Emission Tomography Imaging

Before the PET scan, an magnetic resonance imaging (0.3 T MRP7000AD; Hitachi, Tokyo, Japan) was performed with three-dimensional mode sampling (repetition time (TR)/echo time (TE): 200 milliseconds/23 milliseconds, 75° flip angle, 2 mm slice thickness with no gap, 256 × 256 matrices) for setting regions of interest (ROIs). As described elsewhere, we were able to place ROIs on the region in the reconstructed PET images parallel to the AC–PC line without reslicing (Ouchi et al, 1998).

The PET procedure has been described in detail elsewhere (Klumpers et al, 2008; Kono et al, 2010; Ouchi et al, 2002). In brief, we used a high-resolution brain PET scanner (SHR12000; Hamamatsu Photonics KK, Hamamatsu, Japan: intrinsic resolution, 2.9 × 2.9 × 3.4 full-width at half-maximum, 163 mm axial field of view) with 24 detector rings yielding 47-slice images simultaneously. After head fixation using a thermoplastic facemask and a 10-minute transmission scan for attenuation correction, serial scans (time frames: 4 × 30 seconds, 20 × 60 seconds, and 8 × 300 seconds) were performed for 62 minutes after a slow bolus venous injection of a 6-MBq/kg dose of the tracer. Arterial blood sampling was not performed in the present study. During PET measurements, all patients were monitored online to determine whether any motor movement was present as a symptom of RLS. As we have reported elsewhere, neither emotional nor motor stimulation during the [¹¹C]raclopride PET study was used to verify that the measurement of [¹¹C]raclopride binding could reflect resting state values (Ouchi et al, 2002, 2007).

Image Data Processing

Irregular ROIs were placed bilaterally over the NA, caudate, ventral and dorsal putamen, and cerebellum on the magnetic resonance images (Mai et al, 1997; Yagi et al, 2010). The ROIs were then automatically transferred onto corresponding PET distribution images using image processing software (Dr View, Asahi Kasei, Tokyo, Japan) on a SUN workstation (Ultrasparc, SUN Microsystems, San Diego, CA, USA) (Ouchi et al, 1999b). As described previously (Klumpers et al, 2008; Ouchi et al, 2002), the cerebellum was selected as the reference region for [¹¹C]raclopride. Using time-activity curves of target and reference regions, the BP_ND in each region was estimated with a noninvasive Logan plot analysis (Logan et al, 1994). Because the PET distribution image used for the ROI setting was composed of two consecutive images that covered slices of 6.8 mm in thickness in the z direction (Ouchi et al, 1998), each ROI value contained functional information about the striatum to a depth of at least 6.8 mm in the z direction (volume data). In addition to ROI analysis, to perform voxel-based analysis, BP_ND parametric maps were also generated using voxel-wise Logan analysis implemented in the PMOD software (PMOD 3.0, PMOD Technologies, Zurich, Switzerland). Parametric maps were normalized to the Montreal Neurologic Institute standard brain using statistical parametric mapping software (SPM8; http://www.fil.ion.ucl.ac.uk/spm). After smoothing images with an isotropic Gaussian kernel of 6 mm full-width at half-maximum, t-statistic analysis was performed on a voxel-by-voxel basis without grand mean scaling for contrasts of the between-group condition, resulting in t-statistic maps (SPM{t}). Subsequently, these maps were transformed to a unit normal distribution (SPM{Z}).

Statistics

Changes in clinical scores (IRLS, ESS, and PSQI) were analyzed by the Wilcoxon signed-rank test. In ROI analysis, a one-way analysis of variance was performed to evaluate all estimates in the brain regions between groups with Bonferroni's test for the correction of multiple comparisons. Significance was set at P<0.05 since post hoc multiple comparisons were performed. Although there was no significant difference in age between groups, multiple regression analyses were performed to compare clinical scores with tracer binding in each region under the condition that age was treated as a nuisance variable. In these correlation analyses, significance was set at P<0.05. In SPM analysis, significance was also set at P<0.05 with family-wise error correction (FWE) for multiple comparisons.

Results

Changes in Clinical Scores

As described in Figure 1, the mean ILRS score in the patient group was significantly lower at 1 day after initiation of treatment, and continued to decrease during the subsequent 2 weeks of pramipexole therapy. Similarly, the PSQI was found to be significantly different at day 3 posttreatment and decreased thereafter. These changes indicated an acute efficacy of pramipexole for the RLS symptoms. The ESS value did not vary during the follow-up period, indicating that the present low dose of pramipexole hardly affected daytime sleepiness.

Figure 1.

Changes in clinical scores after D2/3 stimulation by pramipexole. There was clinical improvement in the severity score (International Restless Legs Syndrome (IRLS)) and sleep disturbance score (Pittsburgh Sleep Quality Index (PSQI)) at an early stage.

Changes in [¹¹C]raclopride Binding Potential

Patients had no significant movements or urge-to-move sensations in their limbs during PET scans. As described in ROI analysis results (Table 2), there were significant reductions in [¹¹C]raclopride BP_ND in the NA and caudate and a tendency for its reduction in the putamen in the iRLS group. Voxel-based analysis (SPM) also confirmed these results, showing that [¹¹C]raclopride BP_ND levels in the bilateral NA were significantly lower in the disease group (Figure 2; Table 2).

Table 2. Levels of [¹¹C]raclopride binding potential and SPM results.

Brain region	[11C]raclopride BPND		SPM resulta
	iRLS	Normal	Coordinates (x, y, z)	Z score
Mosolimbic dopamine region
Nucleus accumbens	1.14 (0.21)b	1.37 (0.47)	Right: 18, 20, −4	5.05
			Left: −12, 10, −2	4.62
Caudate	1.41 (0.37)b	2.12 (0.21)
Nigrostriatal dopamine region
Ventral putamen	2.02 (0.15)	2.67 (0.21)	Left: −20, −18, −2	4.96
Dorsal putamen	1.70 (0.26)	2.37 (0.35)

BP, binding potential; iRLS, idiopathic restless legs syndrome; SPM, statistical parametric mapping.

Data shown are mean values (s.d.).

^aOnly brain areas with P value <0.05 corrected for multiple correction (FWE) are described.

^bP<0.05 versus normal control (Student–Newman–Keuls test).

Figure 2.

Significant reduction in [¹¹C]raclopride binding potential (BP_ND) in idiopathic restless legs syndrome (iRLS) patients. [¹¹C]raclopride binding decreases in the nucleus accumbens (NA) bilaterally. Color bar indicates a t-value. The coordinates of statistically significant regions are listed in Table 2.

Relationship Between [¹¹C]raclopride Binding and Clinical Scores

There were significant negative correlations of [¹¹C]raclopride BP_ND in the NA with IRLS scores (Figure 3A; y=−0.015x +1.520, r=0.739, P<0.024) and the PSQI scores (Figure 3B; y=−0.031x +1.450, r=0.835, P=0.016). The caudate [¹¹C]raclopride BP_ND tended to correlate with PSQI score (P=0.054). In contrast, no significant correlation was observed in the nigrostriatal projection regions (Figures 3C and 3D).

Figure 3.

Correlations between [¹¹C]raclopride binding potential (BP_ND) and clinical scores in the idiopathic restless legs syndrome (iRLS) group. Significant correlations were observed between levels of [¹¹C]raclopride BP_ND in the nucleus accumbens (NA) (•) and International Restless Legs Syndrome (IRLS) at entry (A), and between NA (•) plus caudate (○) BP_ND and Pittsburgh Sleep Quality Index (PSQI) at entry (B) all in the mesolimbic dopamine projection region. No significant correlation was observed in the ventral () and dorsal (⋄) putamen (C, D).

Relationship Between [¹¹C]raclopride Binding and the Clinical Outcome After D2/3 Stimulation

As shown in Table 1 and Figure 1, pramipexole treatment ameliorated IRLS symptoms, ranging from 14% to 30% changes in recovery, resulting in improvements relative to baseline scores. This posttherapeutic change was positively correlated with [¹¹C]raclopride BP_ND levels in the NA (Figure 4A; y=0.032x +1.476, r=0.872, P<0.05) in iRLS patients. Although there was a tendency for a similar correlation in other regions, [¹¹C]raclopride binding failed to reach a significant level.

Figure 4.

Correlations between the degree of clinical recovery and original levels of [¹¹C]raclopride binding potential (BP_ND). The nucleus accumbens (NA) [¹¹C]raclopride BP_ND was significantly correlated with magnitude of International Restless Legs Syndrome (IRLS) reduction (clinical improvement) (• in panel A). The [¹¹C]raclopride binding failed to reach a significant correlation level in the caudate (○), ventral () and dorsal (⋄) putamen (B).

Discussion

In the present study, we showed significantly lower D2/3 receptor binding in the ventral striatal regions in iRLS patients, suggesting that there is an impairment in D2/3 function at baseline and that the current finding may provide a rational basis for treatment of iRLS with D2/3 agonists. The lower availability of D2/3 receptors was consistent with a recent postmortem study showing that D2/3 receptor density was significantly lower in the striatum of RLS brains (Connor et al, 2009). In the current study, we observed a heterogeneous pattern of D2/3 receptor availability, that is, a reduction in the mesolimbocortical projection region (NA and caudate) and no change in the nigrostriatal region (ventral and dorsal putamen). This finding may highlight one of the advantages of using in vivo imaging techniques on the living brain. The previous postmortem and current antemortem observations suggest that D2/3 receptor impairment in iRLS patients is progressive during their lifetime, but this speculation might best be investigated in longitudinal PET studies.

In the current study, we selected only drug-naive patients who met the iRLS criteria because centrally acting drugs may potentially alter in vivo tracer binding using PET. Indeed, one study showing a significant increase in D2/3 receptor binding (Cervenka et al, 2006), while being otherwise well designed, was conducted in patients who were heterogeneous regarding medication and genetics. In the present study, half of the patients underwent PET scans in the morning and the other half were scanned in the afternoon. Although limited by the small number of subjects, t-statistics revealed no significant difference between patients examined in the morning and those in the afternoon, consistent with an early report showing only a small diurnal effect on [¹¹C]raclopride binding (Cervenka et al, 2008). In addition, monitoring of patients during the scanning session allowed us to exclude the possibility that stress may have been a factor influencing the PET end point, as shown in our previous study (Ouchi et al, 2002).

The decrease in [¹¹C]raclopride binding can be interpreted two ways: one is an increase in endogenous dopamine release in the synapse and the other is a decrease in receptor density due to internalization, degeneration, and/or functional failure. If presynaptic dopamine is impaired, as shown in Parkinson's disease at rest (Ouchi et al, 1999a), a decrease in dopamine release eventually elevates [¹¹C]raclopride binding postsynaptically. Intact dopamine neurons in iRLS (Pittock et al, 2004) are unlikely to generate a large amount of dopamine release in the synapse, resulting in a reduction in [¹¹C]raclopride binding. As shown in our previous study where intact dopamine transporter binding by [¹¹C]CFT and reductions in [¹¹C]raclopride binding occurred in the striatum of idiopathic normal pressure hydrocephalus (Ouchi et al, 2007), the pathological entity may be on the postsynaptic side where the D2/3 receptor can be selectively affected. To confirm this, concomitant use of presynaptic and postsynaptic radiotracers in iRLS patients would have been desirable, as shown in our previous study (Ouchi et al, 1999a). One possible explanation may be proposed on the basis of altered iron metabolism in the postmortem brain of iRLS patients (Connor et al, 2011). However, alterations in iron metabolism in dopaminergic cells may not be a major cause of iRLS because there is no cell loss in the substantia nigra (Connor et al, 2003; Pittock et al, 2004) or area A11 (hypothalamo-spinal dopaminergic projection origin) in iRLS patients.

Although there is no clear answer as to why the D2/3 receptor is significantly affected in iRLS, iron deficiency is an important factor in its etiology. A previous biochemical study showed that the D2/3 receptor expression in cultured cells decreased in an iron level-dependent manner and that this reduction in D2/3 protein density was restored within a few weeks by iron repletion (Unger et al, 2008). In addition, iron deficiency in rats yields a heterogeneous effect on D2/3 receptor density in different brain regions, especially the striatum (Erikson et al, 2001). In the current study, although iron content in the blood was within normal limits, the possibility of a low level of iron in the brain cannot be excluded. Indeed, a recent study suggested a dysregulation of iron uptake and storage within brain microvessels in the RLS brain (Connor et al, 2011). In the current study, we found that the NA, a key structure of the dopaminergic mesolimbic system, was the most affected in the dopaminergic subregions (see Figure 2). This suggests that the NA, the dopaminergic center of emotional control, is the most vulnerable among the striata affected in iRLS. Therefore, there is a possibility that this postsynaptic deficiency in iRLS may predict an impairment of reward systems, and that the experience of inner restlessness is linked to motivational state. A previous histochemical study showed that the D3 receptor was abundant in the limbic striatum including the NA and ventral striatum (Gurevich and Joyce, 1999). The D3 receptor is also important in pathophysiology because iron-deficient mice with D3 receptor expression had sensory and motor symptoms similar to RLS patients (Dowling et al, 2011). Since [¹¹C]raclopride is an antagonistic tracer that binds to D2 and D3 receptors with high- and low-affinity sites, the present results suggested that alterations in not only D2 but also D3 receptors have a role in iRLS pathophysiology. However, the use of an agonistic tracer such as [¹¹C](R)-2-CH3O-N-n-propylnorapomorphine ([¹¹C]MNPA) (Kodaka et al, 2011) could be of more help to understand its pathophysiology in terms of potential alterations in density of the high- and low-affinity sites of D2/3 receptors.

In addition to D2/3 receptor hypofunction in the NA, as mentioned above, area A11 may be an important subregion in the dopaminergic region. Area A11 is located in the posterior hypothalamus and sends dopaminergic axons to the spinal cord, which may be partly related to the regulation of RLS symptoms (Ondo et al, 2007; Qu et al, 2006). Although A11 neurons were found to be intact in a RLS postmortem study, changes in distal A11 synapses may occur in iRLS patients (Earley et al, 2009). Because the NA is relatively close to the A11, and also has a dopaminergic reciprocal projection to the A11, it is possible that D2/3 receptor binding reductions in NA is associated with information on A11 receptor function. A finer PET camera could detect some abnormality in D2 and D3 receptors of the spinal cord dopamine system in the future.

The clinico-biomarker relationships between posttreatment clinical changes and D2/3 receptor binding in the NA were intriguing in the present study. In line with a postmortem study (Connor et al, 2009), there was a significant negative correlation of clinical severities scored as IRLS and PSQI with NA [¹¹C]raclopride binding (see Figures 3A and 3B). These two studies suggest that clinical symptoms could be predicted by the level of D2/3 receptor binding or density in the striatum, even though the subregion of the striatum examined is different. Interestingly, in the present study, a greater improvement after pramipexole treatment correlated positively with smaller reductions in [¹¹C]raclopride binding specifically in the NA (see Figure 4A). Given the finding that all dots in the scattergrams tended to be in the same direction as the NA, D2/3 receptor function in the striatum may allude to the posttreatment prognosis in patients. The rapid efficacy of a minimal dose of pramipexole in just 2 weeks (20.9% improvement on average) supports the idea that a D2/3 agonist should be a first-line agent for iRLS treatment, and may suggest that the dosage would be titrated according to the degree of [¹¹C]raclopride binding in the NA. In addition, the fact that pramipexole, having a relatively high potency for the D3 receptor, is effective suggests that the dopaminergic mesolimbic system should be a therapeutic target in iRLS. Furthermore, as D3 receptors are abundant in the ventral striatum including NA (Graff-Guerrero et al, 2009), the effect of pramipexole may be preferably developed as an activator of NA D3 receptors.

In summary, mesolimbic D2/3 receptor function can be a good surrogate marker of iRLS clinical severity and the magnitude of its binding may suggest future clinical responses after D2/3 agonist therapy.

The authors declare no conflict of interest.