What Is the Role of Dopamine Transporter Imaging in Parkinson Prevention Clinical Trials?

Observation 1

There is about a 50% reduction of age-expected SBR in early patients at the cusp of their clinical PD diagnosis.

This is good in the sense it may be possible to find these individuals to try to slow down or even stave off the onset of motor symptoms. However, it is very hard to find these individuals in a way that is efficient and meets cost requirements for a clinical trial.

Observation 2

DaT imaging demonstrates striatal uptake and SBR with left/right striatal asymmetry and within each striata, anterior–posterior asymmetry (Figure 1

• Download figure
• Open in new tab
• Download powerpoint

Figure 1 Signal Loss and Asymmetry

123-I ioflupane SPECT in a healthy volunteer (A) and de novo Parkinson research participant (B) show characteristic left-right and anterior-posterior asymmetry seen in PD. This pattern of loss has bearing on the optimized uses of DaT imaging in clinical trials. DaT = dopamine transporter; PD = Parkinson disease.

The topographical consequence of these asymmetries are striatal subregional SBRs which follow an orderly pattern of diminution that is maintained on serial scans over 4 years or more.

The difference in the density of transporters moving from caudate (anterior) to putamen (posterior) creates striatal subregions that individually may be best suited to different clinical research roles (Figure 2). For example, the contralateral posterior putamen shows the greatest reduction in SBR, making it the most sensitive region for detecting DaT loss, for example, in an at-risk cohort. At the same time, it may not be best for tracking longitudinal changes, as the low count rates make the variance of the change measure higher than in other regions.¹³ The exception to this might be in cellular implantation scenarios, for example, where stem cells placed in putamen would theoretically be more easily visualized than caudate when assessing transplant viability.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2 Baseline Regional Striatal SBR Analysis

Regional striatal subregion mean SBRs (green bars) demonstrate the topographic pattern consequent of the left-right and anterior-posterior differential asymmetry noted in Figure 1. SBR = specific binding ratio.

The fact that the subregional pattern of reduction is maintained over the disease course suggests a model of progression with similar rates of change for the striatal subregions, but different onset times (or a phase shift) from one subregion to another. This is essential, within a single scan, 6 different snapshots of the stage of pathologic change within that individual.

Observation 3

The SBR signal changes over time, about 11% in the first year and 17%–20% after 2 years.

However, the variance in the percent change in SBR is high with a coefficient of variation of about 100%. Hence, it is feasible to measure changes in a cohort population consistent with progressing disease, but it cannot be reliably used as a single scan to provide information about whether a 10% change is different from a 20% in that person.^12,14,–,16 Those factors which affect SBR, unrelated to DaT density, require close attention (Table 1) to control as many sources of variance, both biological and technical, when using 123ioflupane to measure progressive change. To put it in context, the variance characterizing all serial measures in PD clinical trials is high. Among the various putative progression biomarkers, DaT imaging provides the best power for detecting a signal change in progressing Parkinson patient cohort; however, it is not ideal.

Observation 4

The correlation of SBR with motor and clinical scores is highly significant, but with very modest r values.

This is because the correlation depends on when in the disease course the correlations are assessed. Brain imaging biomarkers provide information about a very specific component of the CNS, while clinical outcomes get filtered and processed through a number of steps downstream from striatal DaT density measures. The imaging and clinical measures are complementary, providing data about different parts of the system. We expect that biomarker assessments are not going to march in stride because they have different time windows of onset, change, and quiescence. For example, patients with early Parkinson disease often have unilateral symptoms (H-Y 1), but SBR changes are noted on both sides of the brain with the contralateral side showing more profound deficits.¹⁷ In this case, there is no correlation between the presence of 25%–30% deficit and motor symptoms because there is no motor sequela. In a short period, the ipsilateral side will manifest symptoms and now there is a correlation with the SBR, which grows stronger with time. Rather than focusing on the lack of correlation, it is useful to consider what additional information is available in the differences between biomarkers and clinical and functional outcomes in a more integrative fashion as applied to clinical trials.

What Does the SBR Actually Measure?

Does SBR measure the density of transporters, neuronal loss, the integrity of the synaptic terminal, all/some/none of these? The loss of signal on DaT-SPECT has been looked at in some nonclinical studies assessing in vivo DaT-SPECT and ex vivo or in vitro substantia nigra pars compacta neuron density, DaT density, or other related measures.^18,–,20 For example, Bäck et al.²⁰ demonstrated in a rat lesion model that in vivo DaT-SPECT signals correlated with ex vivo estimates of striatal DaT density and the number of nigral neurons. Human postmortem correlation with DaT-SPECT reflects the challenges in acquiring these data, especially the long intervals between the scan and pathologic assessment of the brain,²¹ low participant numbers, different scan analyses, different neuropathologic outcomes, and moderate inter-rater agreement among neuropathologists hamper clarification of the question of the translational meaning of the SBR.

It is also important to remember other factors unrelated to transporter density influence SBR, both biological and technical; age and sex; pharmacokinetic factors; rate of uptake; metabolism; genetic variants of the DaT allele; drugs that compete for DaT; and technical factors such as the equipment, resolution, the photon density in the image, data reconstruction and filtering, how regions of interest are placed, etc. Given the various ways noise and variance can infiltrate the data, much effort is placed on image acquisition, processing, and analyses standardization.

How Wide Is the Time Window to Make a Preventative Intervention in PD?

As noted above, there is a significant amount of pathologic change occurring in the brain without clinical manifestations. This window of opportunity for optimal intervention is before motor symptoms when more function may be salvaged. The duration of this interval may be estimated from longitudinal DaT-SPECT imaging in progressing patients with PD over time. Figure 3 shows the longitudinal SBRs expressed as a percent of age-expected uptake for all 6 regional striatal subregions in 278 patients with PD. Each region was fit to a monoexponential curve from which the theoretical SBRs were calculated and back extrapolated to 100% age-expected uptake and the years of premotor symptom change estimate are along the x-axis. There is a pattern of nonoverlapping curves generated from the SBRs. The contralateral putamen has the lowest uptake curve and hence the longest time to reach 100% of age-expected binding. This corresponds to about 13 years over which DaT loss occurs, assuming that the exponential fit is a robust model for DaT signal loss at these early time points.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3 Years to 100% Age-Expected SBR by Subregion

Estimate in years of duration of premotor pathologic progression for striatal subregions based on modeling of the exponential curve fit to Ioflupane SPECT SBRs determined a baseline, year 1, 2 and 4 scans. The most affected region, the contralateral putamen, suggests a nearly 13-year period of change where motor function remains intact. SBR = specific binding ratio.

With this time frame in mind, the critical task is to now identify a clinical assessment algorithm that maximizes the yield of scan-positive patients without motor symptoms and determine, if possible, the expected time before motor symptoms evidence.

How Do We Use At-Risk Cohorts to Help Us in Progression Trials?

At-risk cohorts are particularly attractive for preventative clinical trials because of the possibility of developing Parkinson's motor symptoms within the time window of a clinical study. Two genotypes, glucocerebrosidase A (GBA) and leucine-rich repeat kinase 2 (LRRK2), have been evaluated with Ioflupane SPECT imaging at baseline in nonmanifest carriers and manifest patients with PD,²² the latter with some longitudinal DaT imaging. Both GBA and LRRK2, carriers not manifesting typical idiopathic PD motor symptoms, have subtlety elevated motor ratings and lower Montreal Cognitive Assessment scores relative to healthy volunteers. However, there were no differences on the baseline DaT imaging, in fact the GBA carriers showed slightly higher mean striatal SBR. The motor manifesting LRRK2 group had somewhat milder motor symptoms compared with the idiopathic PD group, but for GBA, there was no difference in motor scores among patients with PD . There was slightly less reduction in the mean striatal SBR for both genotypes compared with the PD group. Moreover, when considering the longitudinal DaT-SPECT in GBA and LRRK2, there are roughly equivalent annualized rates of percent change in striatal SBR for both at-risk genotypes compared with spontaneous PD. These are interesting but preliminary findings in an ongoing study that needs additional data to sort out the relevance of these cohorts to preventative trial designs in PD.²³

Other at-risk cohorts are patients with REM sleep behavior disorder (RBD) or olfactory loss.²⁴ Both of these measures enhance the likelihood of finding an abnormal DaT-SPECT scan in the absence of motor symptoms. The overall positivity rate of SPECT scans in these at-risk groups taken together is about 34%, with slightly higher net positive scans in the RBD patients at about 40%. Baseline scans for these at-risk patients follow a similar topographic pattern to patients with idiopathic PD yet are just slightly higher SBRs.

How Can Imaging Be Used in Clinical Trial Design?

Critical to assessing the feasibility of clinical trial design is the determination of how many patients are needed to demonstrate a significant effect of treatment for a specific study design. Sample sizes are indicated for clinical motor ratings, imaging, and CSF data (Table 2) for a hypothetical study, a 2-arm, placebo-controlled trial where there is 50% or 25% slowing of rate of change of the progression marker (power = 0.8 or 0.9, p < 0.05, 2-tailed). The most power is achieved with DaT imaging because, although noisy, there is less variability in these data than in some of the other measures. This also underscores how difficult it is to measure slow change and in particular a slowing of that slow change. Such measurements are highly susceptible to even small differences in the variance.

How Can We Do Better?

There are a number of ways we can improve and optimize Ioflupane SPECT or other imaging biomarkers in support of PD clinical trials. We first need to optimize the algorithms for identifying imminent at-risk converters and fast progressors and do this more efficiently than currently. We can also be more thoughtful in how we apply SPECT measures to take advantage of the regional heterogeneity of DaT binding. Novel voxel-wise analytic strategies and PET versions of dopaminergic or other targets (α-synuclein) might improve the noise characteristics of the imaging outcome measure.²⁵ More sophisticated integration of biomarkers to take advantage of the complementary information they provide should be leveraged to improve study design and implementation. Finally, the correlation of neurohistopathology with imaging measures will help clarify what imaging is measuring and bring improvements in characterizing PD progression using scintigraphic imaging measures.