The role of radiotracer imaging in Parkinson disease

The biologic process.

Each tracer measures one or more aspect of dopaminergic/monoaminergic nerve terminal function. F-dopa PET reflects the uptake and conversion of fluorodopa to fluorodopamine, but this process involves several steps. After IV administration, F-dopa traverses the blood-brain barrier via the large neutral amino acid transporter, is taken up into neurons by an active transport system, and is converted to fluorodopamine by aromatic acid decarboxylase (AADC), which may become the rate-limiting step in dopamine synthesis in dopaminergic and noradrenergic neurons. F-dopa uptake has been correlated with counts of nigral neurons in 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)-treated monkeys as well as levels of dopamine in the striatum.⁴ There is a correlation (r = 0.84) between F-dopa uptake and subsequent nigral cell count in humans, although only one of the seven subjects had PD, and it is in PD that compensatory changes might be most likely to occur that would potentially reduce this correlation.⁵ F-dopa uptake is not thought to be a direct measure of nigral cell count, but a measure of AADC activity in those cells.

The type 2 vesicular monoamine transporter (VMAT2) is the protein responsible for pumping monoamines from the cytosol into synaptic vesicles and is distinct from VMAT1, which is found in the adrenal medulla. DTBZ PET is therefore intended to measure synaptic vesicles containing monoamines (dopamine, serotonin, norepinephrine, histamine).⁶ In the striatum, more than 95% of VMAT2 is associated with dopaminergic terminals.⁷ Newly synthesized as well as recovered dopamine is present in these vesicles. The vesicles prevent catabolism of the neurotransmitters and store them for subsequent exocytotic release into the synaptic cleft. Reduced DTBZ binding has been shown in rodent models of PD⁸ and in PD brains compared with control brains.⁹

There are several dopamine transporter (DAT) markers, but B-CIT SPECT has been used in large clinical trials and is the focus of discussion. DAT is specific to dopaminergic neurons. Dopamine reuptake through the DAT is the primary mechanism of dopamine removal from the region of the synaptic cleft, and thus DAT expression is regulated at several levels.¹⁰ B-CIT, which is a tropane derivative, binds to the DAT and with lower affinity to the serotonin transporter. There is evidence that DAT mirrors the decline in levels of striatal dopamine in MPTP-treated monkeys¹¹ and that DAT expression within dopaminergic neurons is reduced in PD brains compared with controls.^12,13

Nigrostriatal dopaminergic function can be indirectly assessed with metabolic imaging using FDG PET, a marker of resting glucose utilization. Multivariate analysis of regional patterns seen with FDG PET can identify specific metabolic brain networks associated with PD. This is an exploratory approach to analyzing imaging data that has been less widely studied than the other techniques. This technique is based on principal component analysis and allows identification of disease-related patterns of regional metabolic covariation (i.e., brain networks) and quantification of pattern expression (i.e., network activity) in individual subjects.¹⁴ This is not the only way to analyze FDG PET data, but this method yields a PD-related pattern (PDRP) that is characterized by pallidal and thalamic hypermetabolism and metabolic decrements in the lateral premotor cortex, the supplementary motor area, the dorsolateral prefrontal cortex, and the parietooccipital association regions. This technique, unlike the three ligands, does not assess specific aspects of dopaminergic neurons but rather network activity related to dopaminergic function. Expression of the PDRP correlates with F-dopa PET¹⁵ and is modulated by dopaminergic therapy.¹⁶

Each of these neuroimaging techniques measures one or more functions of dopaminergic neurons. Although the precise causes of sporadic PD remain unknown, the nigrostriatal dopaminergic system has been the traditional focus of both diagnostic criteria and therapeutic interventions in PD. The face validity of current radiotracers supports their use as biomarkers. However, these techniques do not assess the number or density of nigral dopaminergic neurons, and measurements derived from these tracer uptake studies do not directly measure the biologic processes under study in the same way as in vitro studies using tissue samples. All these imaging methods require simplifying assumptions for data acquisition and analysis and can be affected by factors other than the primary biologic process under study. Therefore, the correspondence between tracer uptake and tissue biology will always be imperfect.

Loss of nigrostriatal dopaminergic neurons and changes in their function is central to PD, but nigral cell loss is just one of several steps on the pathway to clinical symptoms. At the proximal end of the pathway are the suspected inciting and perpetuating causes of PD, such as oxidative stress, mitochondrial dysfunction, proteasome dysfunction, and protein aggregation.¹⁷ These processes eventually lead to nigrostriatal dysfunction and cell death. Braak et al.¹⁸ have inferred an ascending pattern to the evolution of PD-related pathology. Stages 1 and 2 involve structures of the olfactory system and brainstem nuclei (the dorsal IX/X motor nucleus and reticular zone and caudal raphe and locus ceruleus complex), stage 3 involves midbrain and nigral pathology, and stages 4 and 5 with 6 show expanding areas of cortical involvement. Early and later stages involve primarily nondopaminergic systems. Predominantly nondopaminergic symptoms such as depression, cognitive impairment, and postural instability are major contributors to disability in PD^19,20 but may not be captured by dopamine-related tracers. Dopamine trafficking in nigral cells is an important component of PD and may reflect or parallel the burden of disease in other areas of the brain, but the temporal evolution of PD suggests a more complicated relationship of different brain regions to clinical symptoms.²¹

1) Explore biology and evolution of disease.

The three dopaminergic ligands have been evaluated in several natural history studies. These studies demonstrate the evolution of nigrostriatal dopaminergic function and its relationship to the clinical features of PD. For this discussion, we consider natural history studies to be either cross-sectional or longitudinal studies in any stage of PD, including treated natural history studies but not those involving experimental interventions.

Two dopaminergic techniques show statistically significant negative correlations of striatal uptake with motor function as assessed by the Unified Parkinson’s Disease Rating Scale (UPDRS) in the practically defined “off” state: F-dopa, r = −0.68²² and B-CIT, r = −0.48 to −0.81 (a negative correlation because higher UPDRS scores reflect more severe PD and correspond to lower uptake).^23,24 DTBZ binding is not significantly correlated with “off” UPDRS scores but is significantly correlated with duration of PD (r = 0.53) and Schwab and England Activities of Daily Living scales (r = 0.38) (Roger Albin, personal communication, 2003).

In longitudinal studies, F-dopa PET and B-CIT SPECT show a 4% to 13% yearly reduction in baseline putamen uptake compared with 0 to 2.5% in healthy controls.^25,26 Extrapolations from linear regression analysis of F-dopa PET imaging have been used to estimate the duration of the preclinical period of PD.^27,28 For DTBZ, one study has shown a 6.3% change in PD versus a 1.7% change in controls over 2 years (Roger Albin, personal communication, 2003). Importantly, change over time in these studies is expressed as a percentage from an already reduced baseline. Thus, the absolute change compared with normal would be smaller. The estimated correlation coefficient between PDRP network expression and the “off” state UPDRS is approximately 0.8.^16,29

Natural history studies for each radiotracer have had small sample sizes, but the results are consistent. This suggests that these tracers measure biologic processes that are related to the duration and severity of PD as measured by motor rating scales in the practically defined “off” state. Each of these techniques measures an aspect of dopaminergic function, none of which individually conveys the overall function of dopaminergic neurons. In fact, a study comparing F-dopa, a DAT ligand, and DTBZ in the same PD subjects showed evidence of relative upregulation of AADC and downregulation of DAT. These findings are consistent with the idea that surviving neurons would synthesize more dopamine and take up less dopamine from the synaptic cleft.³⁰

2) Studies of diagnosis and prognosis.

For diagnostic or prognostic use, a biomarker must determine true disease status or predict an important clinical outcome. Diagnostic studies performed with the four tracers have generally shown differences between groups of people with PD compared with healthy controls or patients with essential tremor (ET). The ability to distinguish groups of typical PD subjects versus normal controls has been shown for FDG PET,³¹ DTBZ PET,³² F-dopa PET,³³ and B-CIT SPECT.³⁴ Studies of B-CIT SPECT with blinded readings report diagnostic sensitivity of greater than 95% and specificity of 83% to 100% for clinically probable PD compared with ET.^34–36 None of these techniques reliably distinguishes idiopathic PD from multiple system atrophy (MSA) or other forms of atypical PD. Preliminary data suggest that FDG PET can potentially distinguish these groups using a discriminant function analysis,^37,38 and this approach is being validated prospectively (David Eidelberg, personal communication, 2003). Additionally, some groups have reported the ability to discriminate PD from MSA and related disorders based on diminished striatal D2 receptor binding.³⁹

Most diagnostic studies have been conducted in small, single-center studies, with variable procedures for blinding readings of the scans. These studies have evaluated only clinically probable cases of PD and used clinical examinations as the gold standard. Few studies include cases in which there was true diagnostic uncertainty to see whether imaging predicts subsequent clinical diagnosis. A recent study showed that quantitative analysis of imaging with B-CIT and SPECT was able to identify subjects with Parkinson syndromes (PD, MSA, progressive supranuclear palsy) in whom there was initial diagnostic uncertainty on the part of the primary neurologist. When compared with the judgment of an expert movement disorders clinician 6 months after the initial visit, imaging showed 92% sensitivity and 100% specificity for Parkinson syndromes in 35 subjects.⁴⁰ Overall, however, the application of these tests in the setting of diagnostic uncertainty is not known. Distinguishing groups, such as PD and ET, is an early demonstration of the utility of a diagnostic test. Further studies must show adequate sensitivity, specificity, positive and negative predictive values, and likelihood ratios. These studies should be conducted in the target population in which the test is intended to be used and must show that the clinical application of the test is useful in changing the care of patients and/or reducing costs.

Clinical experience suggests that judicious use of DAT imaging in the setting of diagnostic uncertainty may be useful in the hands of the movement disorders specialist, but this has yet to be confirmed by a prospective study. Although there is increasing use of diagnostic imaging by investigators, currently it is not recommended that DAT and other tracers be used routinely in the diagnosis of PD or other parkinsonian syndromes.

3) Therapy development and evaluation.

In early drug development, RTI may be used to assess blood-brain barrier penetration, dose-related occupancy at the site of interest, selectivity for the target, and distribution and kinetics of the compound. Responsiveness to or regulation by dopaminergic agents would be an asset in answering these types of questions, and RTI might assist in the selection of active compounds for further clinical studies. The four tracers described here have not been used for early drug development in this way. However, the potential is clear and there are examples relevant to PD, such as the D2 ligand [¹¹C]raclopride that has been used in developing atypical neuroleptics.⁴¹

Several Phase III and IV clinical trials have used imaging with F-dopa PET or B-CIT SPECT as a primary or secondary outcome (table 1). These clinical trials demonstrate the uses of imaging in the later phases of therapy development. They also show the relationship of RTI to CEs in different patient populations and with different interventions.

The Elldopa trial randomized 361 patients with early PD (<2 years of PD and not requiring symptomatic therapy) to placebo or one of three doses of carbidopa/levodopa: 12/50 mg TID, 25/100 mg TID, or 50/200 mg TID.⁴² Subjects were followed for 40 weeks on the intervention and then began a 2-week washout phase. The primary endpoint was change in “off” motor scores from baseline to week 42, with the hypothesis that levodopa might hasten the underlying progression of PD despite its control of symptoms. Subjects underwent B-CIT SPECT at baseline and week 40. All three levodopa treated arms were significantly better than placebo on the primary outcome, pre- and post-washout, although the long duration of effect of levodopa may account for some of the observed differences. Despite the clinical benefits seen with levodopa, RTI showed the opposite: a reduced rate of decline in striatal tracer uptake in the placebo group (−1.4%) compared with the three levodopa groups (−4% to −7.4%, p = 0.035).

Two clinical trials using RTI evaluated the effects of initial dopamine agonist administration versus levodopa in early PD on motor complications, UPDRS scores, and quality of life. One trial evaluated initial pramipexole versus levodopa (CALM-PD).⁴³ In this study, a subset of patients received scans. The other trial evaluated initial ropinirole versus levodopa and all subjects received imaging (REAL-PET).⁴⁴ The interpretation of the results is debated, but both trials showed that agonist-treated subjects were less likely to experience motor complications than levodopa-treated subjects.^45,46 In the CALM-PD study, UPDRS scores for the full cohort were better in the levodopa arm at 24⁴³ and 48⁴⁵ months. In the imaging substudy, there was no difference in “off” motor and total UPDRS scores between treatment arms at 46 months.⁴⁵ In the REAL-PET study, levodopa-treated subjects had better controlled PD than agonist-treated subjects as measured by “on” scores of the UPDRS motor section but similar Global Clinical Impressions scale scores. In both studies, subjects randomized to initial levodopa had less somnolence and edema.

In terms of the RTI outcomes, both agonist trials showed similar results, but at different time points. The pramipexole versus levodopa trial used B-CIT SPECT. At 24 months, there was no significant difference between the levodopa- and agonist-treated groups.⁴³ After 46 months, there was a significant one-third reduction in the relative rates of decline in RTI in the agonist-treated arm compared with the levodopa-treated arm.⁴⁶ The ropinirole study used F-dopa PET and had a similar finding of a one-third relative reduction in the rate of decline in the agonist-treated group at 24 months.⁴⁴ Thus, RTI was consistent with aggregate measures of motor complications but not other measures of PD severity, quality of life, and global function.

The interpretation of imaging data from these clinical trials is challenging because of the potential for direct pharmacologic regulation of the targets of these ligands.^47–49 Levodopa and dopamine agonists have the potential to alter imaging by changing the uptake or metabolism of F-dopa or by changing the number, occupancy, or affinity of receptors that bind DAT tracers. These direct effects on tracers may be separate from effects on disease progression. The duration of these pharmacodynamic effects is often unknown, making washout designs problematic.^47–49

Studies in patients with PD designed to assess direct pharmacologic effects of medications on RTI targets are characterized by varied interventions and measures, small sample sizes, and limited power to detect effect sizes that could be significant in larger clinical trials. For example, there are five published studies of direct drug effects on DAT ligands (table 2). Three studies showed no statistically significant change on short-term imaging measures with levodopa or dopamine agonists.^46,50,51 One showed a nonsignificant upregulation of striatal DAT by dopamine agonist administration.⁵² Another study using the DAT ligand [¹¹C]RTI-32 and PET showed a significant reduction in DAT expression in all striatal regions in the levodopa group and smaller but significant reductions in the pramipexole and placebo groups.⁵³ Although a different tracer was used, the findings with [¹¹C]RTI-32 and PET may explain the results of the Elldopa trial and CALM-PD trials. The varied results from these five small studies do not adequately answer questions about direct effects of drugs on tracers, and they underscore the complexity of this issue with multiple possible permutations of drugs, doses, and tracers over variable periods of study.

F-dopa PET has been used in tissue transplantation trials. In these trials, RTI provided information about graft survival and function. In the first double-blind, placebo-controlled surgical trial of human embryonic dopaminergic tissue transplantation, 40 subjects with advanced PD underwent F-dopa PET at baseline and 12 months. The primary clinical outcome of this trial was patient-rated global clinical impression of change at 12 months. This outcome was not significantly different between the transplantation and sham surgery groups.⁵⁴ The secondary outcomes of F-dopa PET and “off” scores for the total UPDRS were both improved in the surgical group, with a 40% increase in putamen F-dopa uptake suggesting graft survival and function as dopaminergic cells.⁵⁵ Similarly, a second transplantation trial showed a nonsignificant trend to improvement on the primary outcome of “off” UPDRS motor scores but significant improvement in F-dopa PET.⁵⁶

4) SEs in clinical trials.

When used as an SE, a biomarker should predict the important intended and unintended (e.g., severe expected and unexpected adverse events) clinical results of an intervention with a high degree of accuracy. Although there are moderate correlations of RTI with clinical rating scales, these do not imply a predictive relationship, and correlations may oversimplify the relationship between CE and RTI.⁵⁷ The clinical trials described here highlight the variable relationship between RTI measures and CEs and show why no RTI technique should be considered an SE in PD at this time. In fact, no clinical trial in PD has used an RTI measure as an SE. All efficacy studies have had clinical primary outcome measures with the exception of the REAL-PET study, which used F-dopa PET as the primary outcome measure but otherwise replicated previous longer and larger studies with primary clinical outcomes.⁵⁸ This Phase IV trial was not designed to lead to drug approval or a change in indications or labeling, and thus it did not truly use F-dopa PET as a surrogate.

5) Assisting in the choice of and monitoring response to therapy.

Therapeutic monitoring requires that the biomarker either function as a prognostic tool or reflect information about the risks or benefits of treatment. As with surrogates, the data to support RTI in this role do not yet exist and RTI is not currently used in this manner.