Imaging dopamine system integrity

Perhaps the most explored functional imaging area in PD is that addressing dopamine neurotransmission integrity. This reflects the importance of dopaminergic dysfunction underlying PD symptoms, symptom response to dopaminergic therapies and the classic nigrostriatal pathological features identified in PD [2]. These imaging techniques provide the ability to assess pre-, intra- and postsynaptic dopaminergic transmission [3, 4]. Presynaptic dopaminergic imaging relies primarily on positron emission tomography (PET)- and single-photon-emission computed tomography (SPECT)-based modalities: levodopa (l-DOPA) that has been radiolabeled (¹⁸F-DOPA) is taken up and processed presynaptically; radiolabeled dihydrotetrabenazine binds vesicular monoamine transporter type 2 (VMAT2), important for presynaptic vesicle processing; and radiolabeled cocaine-like agents (tropanes) bind to the dopamine transporter, which is located presynaptically and is important in clearance of dopamine from the synaptic cleft [5]. Presynaptic radiotracers have been shown to have reduced uptake in the posterior more than the anterior striatum, correlate with nigrostriatal pathological features and correlate with bradykinesia (but not tremor) in PD [6]. Despite these clinical and pathological correlates with PD, changes in presynaptic cellular mechanisms (pathological or compensatory) may impact measured outcomes and there are inconsistent findings regarding clinical and pathological progression relative to radiotracer-based imaging [7]. Furthermore, subjects with clinical parkinsonism but without evidence of dopaminergic deficiency (known as SWEDDs [scans without evidence of dopamine deficits]) as measured by such imaging techniques represent another challenge in correlating dopaminergic imaging output with clinical syndrome [8].

Intra- and postsynaptic dopaminergic imaging techniques also rely primarily on PET and SPECT [9], including the ability to secondarily detect changes in synaptic cleft dopamine levels as indicated by alterations in D2-receptor availability for radiolabeled agents with low binding affinity (e.g. ¹¹C-raclopride). Changes in synaptic dopamine as measured by these techniques have been correlated with clinical and pathological features in PD, but this is complicated by the difficulty in accounting for the impact of receptor internalization and endogenous dopamine processing and by observations that certain clinical features do not correlate with these imaging measures [10]. Changes in postsynaptic dopamine-receptor levels have also been evaluated in PD, including the employment of several radiolabeled agents capable of binding D1, D2 and D3 receptors both within and outside the striatum. The interpretation of these data faces challenges similar to those mentioned for intrasynaptic imaging.

Structural imaging

Structural imaging in PD is dominated by MRI, which has permitted investigation of brain functional pathology from micro- to macroscales [11]. Changes in brain microstructure in PD as identified by MRI include abnormalities in substantia nigra morphology identified with high-field magnetic imaging [12], as well as changes in white and gray matter integrity identified by diffusion-tensor imaging. Whole-brain and brain subregion changes have been identified in PD with MRI volumetric studies, and changes in local brain chemistry in PD can be identified with magnetic resonance spectroscopy. Most structural imaging studies in PD to date have focused on correlating imaging findings with diagnosis or symptoms with varying results, with particular attention paid to discrimination of PD versus atypical parkinsonisms (e.g. multiple-system atrophy). Prospective structural imaging studies utilizing MRI in PD are lacking, as are consistent results from large, well-characterized cohorts in cross-sectional studies [13]. Nevertheless, the potential utility of structural MRI in PD was evident in a study incorporating multiple structural measures simultaneously, which reported high accuracy in distinguishing PD from controls when combining three different measurements [14].

Brain-network imaging

There has been a rapid recent expansion in reports of brain-network investigations in PD. The majority of these studies have employed PET-based evaluation of blood flow (H₂¹⁵O-PET) and glucose metabolism (FDG-PET) to investigate brain activity and putative functional connectivity between brain regions, while the potential utility of functional MRI (fMRI) has been explored more recently. Analyses of spatial covariance patterns of PET-derived cerebral metabolism or blood flow measures have reported diagnostic capacity, correlation with symptom manifestation and response to therapy in PD as well [15]. In addition, some PET-based investigations have employed fully automated analytic paradigms, which avoid a priori assumptions regarding connectivity, investigate whole-brain-network function and address potential confounds associated with data acquisition and analysis in PD. Functional MRI-based investigations have also employed network analyses, with particular focus on “activation” patterns and resting-state putative functional connectivity across cortical and subcortical brain regions [11, 16]. These studies have typically used smaller numbers of participants and have yielded less consistent results between studies to date compared with PET-based studies [17, 18], and such techniques must currently be viewed as more speculative.

Other imaging methodologies of interest

Transcranial ultrasound is an inexpensive and accessible imaging technique that has demonstrated increased echogenicity of the substantia nigra in PD, but this finding has also been noted in other parkinsonisms as well as in nonparkinsonian conditions; it appears to be a static measure and has an unclear relationship to PD pathology, symptomatology and disease progression [19]. Imaging of neuroinflammation utilizes a radiolabeled ligand that binds to peripheral benzodiazepine receptors expressed in active microglia, and this technique has been used to demonstrate putative inflammation in the PD brain [20]. There are limited reports using this technique in PD, however, and its relationship to underlying PD pathology, symptomatology and progression is currently unclear. Imaging of neurotransmission beyond dopamine has also been reported in PD [4], with a consistently demonstrated relationship between cognitive impairment in PD and PET-based imaging of reductions in acetylcholine receptor-binding and acetylcholinesterase activity in particular [21]. Imaging the core pathologies of PD (e.g. α-synuclein aggregation) has proved challenging, but recent advances in this area provide reason for optimism for future development of such markers [22].

Multimodal imaging

Given the wide range of potential targets for functional imaging in PD, in combination with the broad pathology, pathophysiology, symptomatology and patterns of PD presentation and progression, a multimodal approach could be important in developing PD FIMs [23]. Recent reports demonstrate the evolving feasibility of multimodal imaging, including the MRI study mentioned above utilizing multiple MRI-based structural measures simultaneously. The capacity to image neurotransmitter, synaptic, structural and network dynamics contemporaneously in PD using a system that combines the molecular sensitivity, spatial and temporal capabilities of PET and MRI is also tantalizing [24], but currently this area is experimental.

Consideration of functional imaging marker criteria in Parkinson’s disease

Criteria have previously been proposed regarding PD FIM usage to measure disease progression [26] and as a surrogate for a clinical endpoint [1]. In tracking PD progression, suggested FIM criteria included the ability to measure a process that changes with anatomic and symptomatic disease progression, objectivity, reproducibility, specificity, safety and availability. Considering FIM use as a surrogate endpoint for a clinical outcome, suggested biomarker criteria include the ability to measure disease processes, enabling of diagnosis/prognosis, evaluating therapeutic efficacy and mechanism, guiding therapeutic choice, and being tightly correlated with its associated clinical endpoint.

No single currently available functional imaging measure meets all of these criteria. Furthermore, it can be argued that no one FIM could meet all the criteria: given the varied pathological, neurophysiological and clinical PD sequelae, it is difficult to conceive a single imaging measure that could capture all of these signals with equal validity at all stages of disease. In addition, a given PD FIM may have more relevance for one particular disease subtype (e.g. tremor-predominant PD), pathological hallmark, symptom, treatment response or disease complication than for another. Indeed, the same FIM that may tightly correlate with early disease pathology and symptomatology could lack utility as PD progresses, yet could still meet the criteria mentioned above at a given disease stage. Such challenges for an FIM in PD are exemplified by the use of dopaminergic imaging in PD clinical trials to date, and this will be expanded on to further consider the meaning of an FIM in PD.

Discordance between clinical and functional imaging marker outcomes in Parkinson’s disease clinical trials

Several clinical trials in PD have attempted to use radiotracer imaging of dopamine system integrity as a biomarker of disease progression correlated with clinical outcome, yet these trials have demonstrated some discordance in this regard [27]. The CALM-PD (Comparison of the Agonist Pramipexole versus Levodopa on Motor complications of Parkinson’s Disease) trial investigated the effect of pramipexole versus levodopa both clinically and with regard to impact on dopamine transporter (DaT) binding as a presumed measure of dopamine neuron degeneration [28]. The primary outcome was a reduction in loss of DaT binding in the pramipexole cohort, which was discordant with greater clinical improvement in the levodopa group. Similarly, REAL-PET examined the impact of ropinirole versus levodopa clinically and with regard to impact on an ¹⁸F-DOPA uptake measure of dopamine terminal function, again as a presumed marker of neurodegeneration [29]. The primary outcome was a reduction in decreased ¹⁸F-DOPA uptake in the ropinirole cohort, also discordant with observed greater clinical improvement in the levodopa group. Subsequent analyses of data from these clinical trials have suggested that the imaging findings could be explained by differential impact on DaT binding and ¹⁸F-DOPA uptake by levodopa versus pramipexole or ropinirole, respectively, rather than by the reduction in neurodegeneration [7]. Furthermore, similar discordance between clinical improvement and the rate of decrease in DaT binding was noted in ELLDOPA (Early versus Later Levodopa in Parkinson Disease) study, which was designed to investigate the impact of levodopa (including dose) on rate of disease progression in PD [30].

In summary, clinical trials utilizing imaging of dopamine system integrity as presumed outcome measures related to neurodegeneration have yielded results that are not fully reconciled with observed clinical responses to medication. This discordance between imaging and clinical outcomes has been interpreted as a possible result of the direct impact medications have on the dopaminergic system, which in turn impacts the imaging measures themselves and confounds their interpretation. This highlights a fundamental problem in developing FIMs as outcome measures in PD trials: can the direct impact of an experimental therapeutic on an imaging measure be fully known to enable the correct interpretation of an FIM as an outcome metric? If not, how tight does the correlation between an FIM and a clinical outcome then need to be and can this be quantified in a manner useful for data analysis in clinical trials? These and other challenges facing FIMs in PD will become important to explore.