Reproducible detection of nigral iron deposition in two Parkinson’s disease cohorts

Abstract

Background

Previous studies investigating nigral iron accumulation used T₂ or T₂*-weighted contrasts to define the ROIs in the substantia nigra with mixed results. Since these contrasts are not sensitive to neuromelanin, ROIs may have inadvertently missed SNpc. An approach sensitive to neuromelanin should yield consistent results. We examine iron deposition in ROIs derived from neuromelanin-sensitive and T₂*-weighted contrasts, respectively.

Methods

T₁-weighted and multi-echo gradient echo imaging data were obtained in two cohorts. Multi-echo gradient echo imaging data were analyzed using neuromelanin-sensitive SNpc ROIs as well as T₂*-weighted SNr ROIs.

Results

Compared with control subjects, significantly larger R₂* values were seen in SNpc of PD patients in both. Mean R₂* values in SNr of PD patients showed no consistency with one cohort showing a small statistically significant increase while the other cohort exhibited no statistical difference.

Conclusion

Mean R₂* in the SNpc defined by neuromelanin-sensitive MRI is significantly increased in PD.

Introduction

Prior to and following the onset of Parkinson’s Disease (PD) symptoms, there is a progressive loss of melanized neurons in the substantia nigra (SN), a paired midbrain structure located near the red nucleus. SN is comprised of SN pars reticulata (SNr) and the SN pars compacta (SNpc) [1]. In healthy subjects, SNpc contains a dense distribution of neuromelanin containing dopaminergic neurons, and at the time of PD symptom onset, a significant portion of melaninzed neurons in SNpc have been lost [2,3]. Iron deposition occurs in SN concurrent with this loss [4].

MRI techniques, such as transverse relaxation rate (R₂ or R₂*) mapping have been applied to investigate PD-related iron deposition in SN [5–18]. However, no clear consensus in regard to iron deposition, as measured with iron sensitive MRI techniques, has emerged in the literature [19–21]. Increases in transverse relaxation rate [5–12,22] in the SN of PD patients have been reported but other studies have observed no differences in SN transverse relaxation rates [13–18]. This lack of consistency may be due small sample size or placement of SN regions of interest (ROIs). In the aforementioned studies, SN ROIs were defined in T₂-weighted images in the hypointense region between the cerebral peduncle and red nucleus. It is important to note that T₂/T₂*-weighted images not sensitive to neuromelanin and localization of SNpc is difficult in these images [23].

Neuromelanin-sensitive MRI (NM-MRI) generates contrast sensitive to neuromelanin (NM) [24–27] and SN seen in T₂/T₂*-weighted contrasts is spatially incongruent to the NM-MRI SN [28]. Thus, prior attempts to estimate iron deposition from PD using ROIs placed in T₂/T₂*-weighted images may not have been placed in the neuromelanin-rich areas that deteriorate in PD. This lack of robustness in identifying the PD-relevant portions of the SN could explain the variability in findings based on ROIs defined using T₂/T₂*-weighted images in previous studies examining PD-related iron deposition. In this work, we examine the reproducibility of PD effects on R₂* in the NM-MRI defined SNpc as well as SNr, defined in T₂*-weighted images, in different populations.

Materials and Methods

All subjects participating in the study gave informed written consent in accordance with protocols approved by the local institutional review boards. Data from two hospitals were analyzed in this study. Data from Cohort 1, consisting of 69 subjects (37 PD and 32 control), were collected at the Emory Movement Disorders Clinic. Cohort 2 consisted of 91 subjects (43 PD and 48 control) and data were collected at Ruijin Hospital. PD subjects were recruited from local movement disorders clinics, and all PD subjects were clinically diagnosed with PD according to the UK Brain Bank criteria [29]. Control subjects at Emory University were recruited from the Emory Alzheimer’s Disease Research Center normal population. Control individuals at Ruijin Hospital were recruited from the general population. Control participants were excluded from the study if they exhibited: 1) symptoms or signs of secondary or atypical parkinsonism [30]; 2) cognitive abnormalities i.e. had Montreal Cognitive Assessment (MOCA) score of less than 26 at Emory University Hospital or Mini-Mental State Examination (MMSE) score of less than 24 at Ruijin Hospital; 3) a history of territorial ischemic stroke or hemorrhagic stroke, epilepsy, brain tumor, multiple sclerosis, neurodegenerative disease, hydrocephalus, bipolar disorder or schizophrenia; 4) treatment with an antipsychotic or any other dopamine-blocking drug; or 5) any contraindications to MRI imaging.

Demographic information, including gender, age and years of education, was collected for each participant (Table 1). Severity of Parkinsonian symptoms were assessed in PD and control groups using the Unified Parkinsonian Disease Rating Scale (UPDRS)-III motor examination. For each cohort, patients were assessed and underwent imaging in the ON medication state.

Table 1.

Demographic information and clinical characteristics of PD patients and healthy controls in cohorts 1 and 2. Data are presented as mean ± standard deviation unless otherwise noted.

Cohort 1				Cohort 2
Variable	HC (n=32)	PD (n=37)	pValue	Variable	HC (n=46)	PD (n=42)	pValue
Gender (M/F)	12/20	19/18	0.26	Gender (M/F)	21/25	17/25	0.63
Age (yrs)	63.0±9.0	64.5±9.2	0.51	Age (yrs)	63.1±7.2	63.6±7.0	0.74
Education (yrs)	16.7±2.5	16.0±2.6	0.25	Education (yrs)	10.7±3.3	12.0±3.8	0.09
MOCA score	27.7±2.4	28.0±1.8	0.61	MMSE score	28.4±1.2	28.6±1.2	0.42
UPDRS-III score	2.2±3.1	19.8±5.9	2.2 × 10−23	UPDRS-III score	—	22.2±12.4	—
Disease Duration (yrs)	—	4.3±4.1	—	Disease Duration (yrs)	—	6.1±4.6	—

Image Acquisition

Data from the first cohort were acquired on a 3 T MRI scanner (Prisma Fit, Siemens Medical Solutions, Malvern, PA) using a 64-channel receive only coil. Images from a MP-RAGE sequence (echo time (TE)/repetition time (TR)/inversion time=3.02/2600/800 ms, flip angle (FA)=8°, voxel size=0.8×0.8×0.8 mm³) were used for registration from subject space to common space. T₂*-weighted data were collected with a 6 echo 3D gradient recalled echo (GRE) sequence: TE₁/ΔTE/TR=4.92/4.92/50 ms, FOV=220×220 mm², matrix size of 448×336×80, in-plane resolution = 0.49×0.49 mm², slice thickness=1 mm, and GRAPPA acceleration factor=2.

Data from the second cohort were acquired on a GE 3 T MRI scanner (Signa HDxT, GE Medical Systems, Milwaukee, WI) using an 8 channel receive only coil. T₁-weighted structural images were acquired using a 3D fast SPGR sequence with the following parameters: TR/TE=5.529/1.724 ms, acquisition slices=196, matrix=256×256, FOV=256 mm, flip angle=12°, and slice thickness=1 mm. T₂*-weighted data were collected with a sixteen echo GRE 3D sequence: TE₁/ΔTE/TR=2.7/2.9/59.3 ms, FA=12°, FOV=220×220 mm², matrix size=128×128×56, 1.72×1.72 mm² in-plane resolution, slice thickness=1 mm, and ASSET acceleration factor=2.

Substantia Nigra Atlases

Standard space T₂*-weighted hypointense SN (SNr) and SNpc neuromelanin atlases from [31] were used in this study. They were transformed to T₂*-weighted images using a process similar to those outlined in [31,32]. A comparison of the spatial position of the two SN ROIs is shown in Figure 1.

Figure 1.

A comparison of SNpc, outlined in blue, and SNr, and outlined in red, ROIs used in the multicohort analysis. The SNpc and SNr ROIs were derived from neuromelanin-sensitive and T₂*-weighted images, respectively. The ROIs are overlaid on the mean R₂* map derived from controls in cohort 1.

R₂* calculation

R₂* values were estimated using a custom script in MATLAB by fitting a monoexponential model to the GRE images.

$$S_i=S_0\exp \left(-R_2^{*}TE\right)$$ [1]

where S₀ denotes a fitting constant and S_i denotes the signal of a voxel at the ith echo time. Mean R₂* was measured in the SNpc mask as well in the T₂-weighted hypointense SNr mask.

Statistical Analysis

All statistical analyses were performed using IBM SPSS Statistics version 24 (IBM Corporation, Somers, NY, USA). Results are reported as mean±standard deviation. Group R₂* comparisons between PD patients and controls were made using a one tailed t-test. A p-value of 0.05 was considered significant for all statistical tests performed in this work. Receiver operator characteristic (ROC) curves were obtained for SNpc and SNr R₂* values.

Results

Mean R₂* values in SNpc, as defined by neuromelanin-sensitive MRI, were increased in the PD group of both cohorts. For cohort 1, mean SNpc R₂* values were 32.5 s⁻¹ ± 5.6 s⁻¹ and 27.5 s⁻¹±4.3 s⁻¹ (p<10⁻⁴) for PD and control groups, respectively. For the second cohort, mean SNpc R₂* values were 34.3 s⁻¹±4.9 s⁻¹ and 29.5 s⁻¹±4.4 s⁻¹ (p=7×10⁻⁴) for PD and control groups, respectively. Interestingly, conflicting results were observed in mean R₂* values in SNr ROIs derived from the T₂-weighted images showed elevated iron levels in the PD group of cohort 1 (PD: 37.8 s⁻¹±5.4 s⁻¹; control: 35.4 s⁻¹±5.2 s⁻¹; p=0.03), but not in the second cohort (PD: 40.7 s⁻¹ ± 6.0 s⁻¹; control: 39.2 s⁻¹±8.4 s⁻¹; p=0.15). These results are summarized in Figure 2.

Figure 2.

Comparison of SNpc R₂* and SNr R₂* in cohort 1 (shown in A and B, respectively) and of SNpc R₂* and SNr R₂* in cohort 2 (shown in D and E, respectively). Statistically significant differences are seen in mean SNpc R₂* both cohorts. In A, B, D, and E, the box denotes the 25^th and 75^th percentile with the line denoting the median value. ROC analyses of nigral iron deposition are shown for cohort 1 in C and cohort 2 in F, respectively.

Mean SNpc R₂* outperformed mean SNr R₂* as a diagnostic biomarker in both cohorts. The area under the ROC curve (AUC) for mean SNpc R₂* in cohort 1 was 0.730 (standard error (SE)=0.082; 95% confidence interval (CI): 0.613–0.842; p=0.001). The AUC for mean SNr R₂* in cohort 1 was 0.604 (SE=0.067; 95% CI: 0.472–0.735; p=0.131). In cohort 2, the AUC for mean SNpc R₂* was 0.751 (SE=0.055; 95% CI: 0.643–0.860; p<10⁻³), while the AUC for mean SNr R₂* was 0.461 (SE=0.055; 95% CI: 0.332–0.590; p=0.552). ROC curves are shown in Figure 2.

Discussion

The establishment of a reliable neuroimaging biomarker for assessing iron deposition in SN is necessary to assist in the development of disease modifying therapies and to track disease progression. Here, we found the mean R₂* in a standard space SNpc ROI derived from NM-MRI images to be a reproducible biomarker in two separate cohorts. Further supporting the generalizability of these results, these cohorts were drawn from different populations, in the United States and China, respectively. These results suggest that mean SNpc R₂* may be a reliable neuroimaging biomarker for assessing PD-related iron deposition.

In contrast, mean R₂^* in the SNr ROI, as defined in T₂-weighted images, was found to be an inconsistent biomarker with cohort 1 showing a statistically significant difference between groups and cohort 2 not exhibiting a statistically significant difference between groups. This inconsistency in observing PD-related iron deposition using T₂-weighted images to define SN ROIs is in accordance with prior work [5–18]. The inconsistency of iron deposition present in prior studies may be at least partially explained by insensitivity of T₂- and T₂*-weighted images to neuromelanin [33] and to differences in ROIs used for SNpc and SNr in evaluating PD-related SN iron deposition [6,10,34]. In most of the previous studies, the SN ROI was defined to be the hypointense region between the red nucleus and cerebral peduncle in T₂/T₂*-weighted images. As NM-MRI signal colocalizes with neuromelanin [35] and nigral neuromelanin is present exclusively in the SNpc [2], these previous ROIs resided mostly in SNr, as opposed to SNpc, which is the critical structure for PD [36].

A biological rationale for the conflicting results observed in mean SNr R₂* values of the two cohorts examined in this study may be attributed to the spatial locations T₂*-weighted (i.e. SNr) and NM-MRI (i.e. SNpc) SN volumes. Prior work has found a non-zero overlap between the T₂*-weighted (i.e. SNr) and NM-MRI (i.e. SNpc) SN volumes [28]. In some subjects, iron deposition may be so pronounced in this region that it effectively skews R₂* in the rest of the T₂*-weighted SN ROI (i.e. SNr). In this case, a few subjects exhibiting this trait may induce statistically significant changes in studies with small populations.

As reported in histology [2,3], much of the degeneration in SN occurs inferior to the red nucleus. In imaging studies, this degeneration occurs in the superior and lateral-ventral portions of SNpc [37]. Some studies using T₂/T₂*-weighted images to define SN ROIs have placed these ROIs regions of SNc experiencing iron deposition and found increases in free water [38,39]. Studies using T₂/T₂*-weighted should choose ROIs in the inferior and posterior portions of the SN in T₂/T₂*-weighted images, or define SNc ROIs from NM-MRI [24–27] or multi-spectral MRI [40].

Conclusion

In summary, this study shows SNpc, as defined by NM-MRI, is sensitive to PD-related iron deposition with reproducible increased R₂*. In contrast, mean SNr R₂*, ascertain on an ROI defined using T₂*-weighted imaging, was found to be an inconsistent biomarker.