Upper brainstem GABA levels in Parkinson’s disease

Abstract

Objective

The dopaminergic pathology of Parkinson’s disease (PD) impacts circuits involving GABAergic neurons, especially in the brainstem, where the disease manifests early. The aim of this study is to test the hypothesis that levels of gamma-aminobutyric acid (GABA) in the upper brainstem are reduced in patients with PD compared to healthy controls, using edited magnetic resonance spectroscopy (MRS of GABA +).

Materials and methods

GABA + levels were examined in 18 PD patients and 18 age- and sex-matched healthy controls (HCs). GABA + -edited MRS was performed in 7.5-ml voxels in the upper brainstem, and the spectra were processed using the Gannet software. Differences in GABA + levels between the two groups were analyzed using independent t test analysis.

Results

GABA + levels were significantly lower (p < 0.05) in the upper brainstem of the patients with PD (4.57 ± 0.94 mM) than the HCs (5.89 ± 1.16 mM).

Conclusion

The lower GABA + levels in the upper brainstem of the PD patients suggest that a GABAergic deficit in the brainstem may contribute to the pathology in PD.

Introduction

Parkinson’s disease (PD) is a neurodegenerative disorder caused by degeneration of dopaminergic neurons in the substantia nigra (SN) which leads to dysfunctional motor control [1]. Intraneuronal Lewy bodies and Lewy neurites are the pathological hallmarks of PD. The earliest clinical signs of PD are non-motor symptoms (NMS: depression; hallucinations; olfactory disturbance), but diagnosis usually only occurs with the advent of motor symptoms (bradykinesia, akinesia, rigidity, and tremor) which occur following the loss of a large number of dopaminergic neurons.

Many circuits involving dopaminergic projections from the midbrain also include GABAergic inhibitory projections [2, 3]. For example, those to the striatum are mirrored by GABAergic projections back to the midbrain and substantia nigra [2]. Altered dopaminergic activity in PD substantially impacts the GABAergic circuitry of the basal ganglia (BG) [4] and SN [2], and impacts the excitation/inhibition balance in cortex. Lewy bodies, the pathological hallmarks, appear in the brainstem before their emergence in BG and cortex of PD in the majority of cases [5-7]. However, this predominantly upward course has been challenged by a number of recent studies ([8-10], and for review [11]). This motivates our study of GABAergic changes in the dopamine-depleted brainstem, as an important prerequisite for understanding the pathological and neurometabolic response that take place within both the SN and the BG network.

Magnetic resonance spectroscopy (MRS) provides a non-invasive tool for measuring levels of brain metabolites [12]. While dopamine levels are below the MRS-detectable threshold in the brain, GABA can be measured using edited MRS, which might offer a tool for investigating early cellular and metabolic changes in PD. J-difference editing is the most commonly used strategy for MRS of GABA, because it alleviates signal overlap with other metabolites such as glutamate (Glu), N-acetyl aspartate (NAA), and total creatine (Cr), to reveal the signals of low-concentration GABA. J-difference editing with Mescher–Garwood Point Resolved Spectroscopy (MEGA-PRESS) [13] method has been widely used in the brain of normal humans [14-16] and patients with neurodegenerative [17-22] and other disorders [23, 24]. Edited MRS of GABA is a challenging methodology, and more so in the brainstem, due to its small size (making typical cortical measurements volume of 27 ml impossible), irregular shape and deep location. Thus, few studies have sought to measure GABA + levels in the brainstem, and even fewer in PD.

Studies of GABA + in PD have investigated multiple brain regions with a range of cohort sizes, and perhaps unsurprisingly yielded mixed results. In cortical regions, studies have shown increased GABA + (in medial prefrontal cortex [20]), decreased GABA + (in visual cortex [22]) and no change (in motor cortex and visual cortex [19]). Both increased [25-27] and decreased [17, 18] GABA + levels have been reported in the BG, as well as studies reporting no significant difference [21, 28]. In thalamus, both increased GABA + levels [28] and no significant difference [19, 21] in GABA + levels have been reported. In the present study, based on our previous data [17, 18] and GABA-collapse hypothesis proposed by Blaszczyk [29], we hypothesize that GABA + levels (GABA + macromolecules + homocarnosine, as co-edited with MEGA-PRESS) are reduced in PD in the upper brainstem, where pathological changes precede those seen in the basal ganglia.

Materials and methods

Participants

Between April 2018 and April 2019, 18 PD patients (aged 35–75 years) and 18 sex- and age-matched healthy controls (HCs) were admitted to be scanned by 3 T MRI. The study was approved by the local institutional ethical review board, and written informed consent was obtained from each participant. For all participants, exclusion criteria included contraindications for MRI, and history of alcohol or substance misuse. Although all the PD patients were receiving treatment with antiparkinsonian medication, they did not take their medication for at least 12 h prior to MR imaging [30].

MR acquisition

All MRI and MRS experiments were carried out using a 3 T scanner (Achieva TX, Philips, Best, Netherlands) equipped with an eight-channel phased-array head coil.

Structural magnetic resonance imaging data were acquired using a T₁-weighted three-dimensional Magnetization-Prepared Rapid-acquisition Gradient Echo (MPRAGE) sequence with sagittal acquisition, resolution 1.0 × 1.0 × 1.0 mm³; TR/TE 8.2/3.7 ms; flip angle = 8°, matrix = 256 × 256; field of view = 24 × 24 cm².

MEGA-PRESS [13] spectra were acquired from a 1.0 × 2.5 × 3.0 cm³ (AP × LR × HF) voxel centered on the midline of the upper brainstem [31], as shown in Fig. 1. Sequence parameters were as follows: TR/TE = 2000/68 ms; 320 averages (and 8 water reference averages [32]); 2 kHz spectral width; 2048 data points; 14-ms sinc-Gaussian editing pulses applied at either 1.9 ppm (ON) or at 7.46 ppm (OFF) in interleaved fashion; minimum-phase excitation pulse bandwidth 2.2 kHz and slice-selective refocusing bandwidth 1.3 kHz; total scan time = 10 min 56 s. Philips ‘MOIST’ water suppression (bandwidth 140 Hz) and Philips pencil-beam (PB-auto) shimming were used. Since the edited signal detected at 3 ppm using these experimental parameters contains contributions from both macromolecules (MM) and homocarnosine, the signal is labeled GABA + rather than GABA [15].

Fig. 1.

T₁-weighted MRI showing the position of the (10 mm × 25 mm × 30 mm) voxel in the upper brainstem in a volunteer

Three-dimensional T₁-weighted brain images were segmented to calculate the fractional content of grey matter (GM), white matter (WM) and cerebrospinal fluid (CSF) within each spectroscopic voxel using SPM 12 (the Wellcome Institute of Cognitive Neurology, London, UK; fil.ion.ucl.ac.uk/spm/software/). The MRS voxels for each participant were coregistered to the structural image in Gannet [33], and voxel volume fractions of GM/WM/CSF were reported. The concentration of GABA in CSF was assumed to be negligible.

MRS data were analyzed using the Gannet software toolkit, a GABA-MRS analysis tool that is coded within MATLAB (The MathWorks, Inc., Natick, MA, USA) using the Optimization and Statistics toolboxes and distributed as open-source software [33]. GABA + quantification proceeded with: (1) frequency-and-phase alignment of FIDs with spectral registration [34]; (2) subtraction of aligned spectra to produce GABA + -edited difference spectra (Glx (glutamate + glutamine) was also co-edited at 3.75 ppm); (3) applying a Gaussian model to quantify the 3-ppm GABA + peak area (NAA, Cr, and Cho levels were also calculated, based on modeling of the edit-OFF subspectrum); (4) calculating tissue-corrected GABA + levels based on the water reference signal, using literature values for water T₁ and T₂ (brainstem GM: T₁ = 0.798 s, T₂ = 0.0788 s; brainstem WM: T₁ = 0.998 s, T₂ = 0.1159 s; CSF T₁ = 3.817 s, T₂ = 0.503 s [35-37]). MR visibility of the different compartments was assumed to be 0.78, 0.65, and 0.97 for GM, WM and CSF, based on [38, 39].

Clinical assessment

The Non-Motor Symptom Questionnaire (NMSQ) was administered to the patient group to evaluate non-motor symptoms. The NMSQ is a single-page, self-administered 30-item instrument which screens for the presence of NMS [40]. It can be completed by patients in 10–15 min. NMSQ contains nine relevant domains: gastrointestinal tract, urinary tract, cognitive, hallucination/delusions, neuropsychiatric, sexual function, cardiovascular, sleep, miscellaneous disturbances. The Unified Parkinson’s Disease Rating Scale (UPDRS) (Part III) and Hoehn and Yahr (H-Y) Stage were applied to evaluate the disease stage of PD patients when the patients were admitted to the hospital, usually ~ 48 h prior to MRS scanning.

Statistical analysis

The results are presented as mean ± standard deviation (SD) values. The concentrations of metabolites such as GABA + in PD patients and healthy controls were compared using the two-tailed independent samples t test, and differences were considered significant for p < 0.05. A multiple regression model (Enter method) was used to test for the relationship between GABA + and the clinical assessment of non-motor symptoms. Since disease duration [41] and PD subtype [18] predict the symptoms of PD, these were also included as factors in the multivariate regression. Although patients did not take antiparkinsonian medication for at least 12 h prior to MR imaging, we also included levodopa equivalent dose (during typical medication) as a factor in the regression model. Exploratory post hoc comparisons were also performed to investigate the link between GABA + levels and the UPDRS III (motor symptoms), as well as individual domains within the NMSQ. Secondary analyses were also performed for Glx results. Statistical analysis was conducted using the Statistical Package for Social Sciences software (SPSS, Chicago, IL, USA).

Results

The study involved 36 participants: 18 Parkinson’s disease (PD) patients and 18 healthy control (HC) volunteers. The demographic and clinical data summarized in Table 1 show that the PD patients mainly presented mild-to-moderate motor symptoms, mild-to-moderate disease stages), and a mean disease duration of ~ 3 years. The PD patients and the HC group did not differ significantly in age (p = 0.163), or sex (p = 0.317).

Table 1.

Demographic and clinical characteristics (presented as the mean ± SD) of the PD and HC subjects

	PD	HCs	p value
Number	18 (tremor-dominant vs. akinetic-rigid: 10 vs. 8)	18	–
Age (years)	60.7 ± 7.6	56.2 ± 11.0	p = 0.163
Number of females (%)	10 (55.6)	7 (38.9)	–
Number of males (%)	8 (44.4)	11 (61.1)	–
UPDRS	32.1 ± 7.6	–
H-Y	2.2 ± 0.5	–
Duration of illness in years	2.9 ± 1.5	–	–
NMSQ	6.6 ± 3.1	–	–

p values of < 0.05 according to SPSS were considered to indicate statistically significant differences PD Parkinson’s disease, HC healthy control, UPDRS unified Parkinson’s disease rating scale, H-Y Hoehn and Yahr stage, NMSQ non-motor symptom questionnaire

Average ± SD for voxel tissue fractions were 0.93 ± 0.02 (WM), 0.03 ± 0.01 (GM), and 0.04 ± 0.02 (CSF). T tests between PD patients and healthy controls revealed no significant differences in segmentation: p = 0.931 (WM); p = 0.551 (GM); and p = 0.784 (CSF).

GABA + -edited spectra were successfully collected in the upper brainstem of all 36 participants; all spectra are shown in Fig. 2, and those of one HC and one PD patient are compared in Fig. 3 with ON and OFF subspectra to visualize Cr linewidth. GABA + levels were significantly lower in the patients with PD (4.57 ± 0.94 mM) than in the HC volunteers (5.89 ± 1.16 mM; p = 0.007). Indices of data quality, such as the GABA + fitting error (p > 0.05) and linewidth (p = 0.11) were not significantly different between groups (as shown in Fig. 4). Importantly, NAA, Cr, Cho, and Glx levels were not significantly different between PD and HC groups (NAA p = 0.82, Cr p = 0.72, Cho p = 0.11, and Glx p = 0.93) (as shown in Fig. 4).

Fig. 2.

GABA + -edited spectra in the upper brainstem of all 36 participants, showing the intended signal at 3 ppm

Fig. 3.

MEGA-PRESS editing of GABA. Edit-ON, Edit-OFF, and GABA-edited spectra from a HC (left) and a PD patient (right). The GABA-edited signals are overlaid with Gaussian models for fitting. PD Parkinson’s disease, HC healthy control

Fig. 4.

Bar charts of the distributions of GABA + levels, normalized fitting errors, linewidth in Hz, NAA, Cr, Cho, Glx levels. PD Parkinson’s disease, HC healthy control, NAA N-acetyl aspartate, Cr creatine, Cho choline, Glx (glutamate + glutemine)

In the multiple regression analysis of GABA + (or Glx) and non-motor symptoms (NMSQ scores), taking the type of PD (tremor-dominant or akinetic-rigid), disease duration, age, and levodopa equivalent dose into account, we found no correlation between metabolite levels and NMSQ scores (GABA + : adjusted R² = 0.150, p = 0.234; Glx: adjusted R² = 0.030, p = 0.407), or UPDRS III scores (GABA + : adjusted R² = 0.030, p = 0.407; Glx: adjusted R² < 0, p = 0.487). GABA + was not as a significant predictor for either NMSQ scores or UPDRS III scores (NMSQ: standardized β = 0.457, p = 0.141; UPDRS: standardized β = 0.247, p = 0.441), neither was Glx (NMSQ: standardized β = − 0.219, p = 0.421; UPDRS: standardized β = − 0.067, p = 0.808). None of other regressors emerged as a significant predictor for either NMSQ scores or UPDRS III scores (standardized β < 0.5, p > 0.1). Additionally, no correlation was seen between GABA + or Glx and individual domains within the NMSQ (adjusted R² < 0.3 and p > 0.1 for all comparisons).

Discussion

The main finding of this study was a significant reduction in the GABA + levels in the upper brainstem regions of patients with PD compared with the HCs, consistent with our hypothesis. To measure GABA + levels in the upper brainstem could facilitate early detection of GABAergic dysfunction before the appearance of nigrostriatal defects and it may potentially be used as a biomarker of pre-symptomatic PD.

All the PD patients included in our study were receiving treatment with antiparkinsonian medication, although the patients did not take their medication for at least 12 h prior to MR imaging. Although previous studies [42] have suggested that PD medication can affect GABA and Cr levels acutely, little is known about longer term changes and our study cannot disambiguate the chronic impact of medication from disease.

The primary pathology of Parkinson’s disease is the degeneration of dopaminergic neurons, characterized by Lewy-body aggregation of misfolded α-synaptophysin, beginning from the lower brainstem (medulla oblongata) to the upper brainstem (pons, midbrain) and then to the limbic system, visceral motor system, and sensorimotor system [5]. Not all PD cases fit the proposed Braak staging scheme (with an upward disease progression), with clinically asymptomatic individuals and individuals with dementia with Lewy bodies (DLB) as notable exceptions [6, 11]. However, in most PD patients with younger onset and long disease duration, the neuropathological progression is consistent with the stages defined by Braak [5, 6]. While the primary deficit is dopaminergic, the GABAergic and dopaminergic systems are heavily interconnected. Many circuits involving dopaminergic projections from the midbrain and substantia nigra also include GABAergic inhibitory projections [2], and indeed, some dopaminergic terminals even co-release GABA [43]. GABAergic neurons located in the brainstem play a role in regulating motor output, which may be impacted in PD.

Although the brainstem is a challenging region for MRS in general, and edited MRS particularly, this study sits in the context of other previous work applying edited MRS in the brainstem [23, 27, 31, 44]. Our work, applying the acquisition protocol previously developed in healthy controls [31], relies on 3 T J-difference editing to quantify GABA (in common with references [17, 18, 23, 25]), whereas two alternative approaches rely upon higher field strength and linear combination modeling, as in references [27, 44] or two-dimensional spectroscopy [45]. Each approach addresses the challenge of resolving the GABA signals from overlapping signals, either by subtracting them out, improving spectral resolution and modeling them, or spreading them into an additional spectral dimension. This study uses a voxel size of 7.5 ml (which is unusually small for 3 T edited MRS), whereas larger voxel sizes (13.5 ml, 30 ml, 18 ml) were employed in references [23, 25] and [18], respectively. Higher field studies focused on the brainstem have used smaller regions of interest (ROIs) (4.5 ml [27], 2.2 ml and 5.3 ml [44]). This manuscript focuses on PD patients, as in references [25, 27]; the brainstem has also been studied in sleep bruxism (SB) [23] and obstructive sleep apnea (OSA) [45], and is likely to be of interest in many pathologies, given the important role of the brainstem in mediating signaling to the periphery and control of basic physiological function.

While several studies have shown change in GABA + levels in different brain are of PD patients, brainstem GABA + levels remain under-studied. In this study, a relatively small ROI (10 × 25 × 30 mm³) was prescribed to measure GABA + levels in the upper brainstem of PD patients. The observed reduction in GABA + levels would be consistent with either GABAergic neuronal loss in the brainstem, or changes in production, utilization or breakdown. Importantly, there was no significant difference observed in NAA, Cr, Cho, or Glx levels indicating that there is no generalized tissue loss [46]. Our GABA results are not well aligned with the one previous study [27] of brainstem GABA + in PD, which reported significantly elevated levels. This study differed from ours own in terms of GABA quantification approach (linear combination modeling vs. editing), TE, voxel size and location (our ROI being larger and positioned higher in the brainstem). It is not clear whether the different results reflect opposite changes in different locations), or methodological differences. They proposed that elevated GABA levels are most likely a result of alterations in several classes of neurons of the brainstem. It is notable that previous studies of GABA in sleep bruxism (SB) [23] have reported changes in opposite directions with 3 T editing approaches.

Just as brainstem changes appear early in the PD time course, non-motor symptoms also precede motor impairment. The Non-Motor Symptom Questionnaire (NMSQ) [47] is sensitive to these non-specific early changes in PD. The major Non-Motor Symptom (NMS) manifestations of PD include neuropsychiatric symptoms (such as depression, apathy, anxiety), sleep disturbances, autonomic symptoms, gastrointestinal symptoms, sensory symptoms(such as pain, olfactory disturbances, visual dysfunction) and other symptoms (such as fatigue) [47]. We found no correlation between GABA + (or Glx) and individual domains within the NMSQ. These functions relate directly to the structures shown to manifest Lewy-body pathology early in the disease progression—the lower brainstem, the central and peripheral portions of the autonomic nervous system, and in the olfactory bulb [48]. Numerous brainstem nuclei are engaged in fundamental homeostatic mechanisms, including gastrointestinal regulation, pain perception, mood control, and sleep–wake cycles [49], all of which are impacted in PD. Our study did not show a significant relationship between symptoms [motor (UPDRS III) or nonmotor (NMSQ scores)] and GABA + levels (or Glx levels) in upper brainstem of PD patients, suggesting either insufficient statistical power or a lack of an underlying role of GABAergic (or glutamatergic) abnormalities in NM symptoms.

Our study has two limitations. First, the sample size in each group was rather small for a clinical study. Therefore, a larger sample size should be recruited to confirm these preliminary results. Second, the detected GABA + signal using the MEGA-PRESS contains a significant contribution (~ 50%) from co-edited macromolecules (MM) and homocarnosine. Methods for acquiring an edited GABA signal without this contamination are desirable, but suffer from increased sensitivity to motion [32]. Therefore, it is frequently accepted as a limitation of this most commonly applied approach [12].

Conclusion

In conclusion, we confirmed the hypothesis that GABA + levels are reduced in the upper brainstem of patients with PD. The results suggest that GABAergic dysfunction may be implicated in PD pathogenesis, suggesting a potential link between regional GABA + levels and pathology progression in PD patients. The results suggest that it may be possible to detect neurotransmitter changes in the upper brainstem of PD patients early, before the clinical symptoms of disease appear, thus allowing an effective early therapy to be implemented.

Abbreviations

PD

Parkinson’s disease

GABA

Gamma-aminobutyric acid

GABA +

Gamma-aminobutyric acid plus co-edited signals

MRS

Magnetic resonance spectroscopy

NMS

Non-motor symptoms

BG

Basal ganglia

ROI

Region of interest

NAA

N-acetyl aspartate

Cr

Creatine

Cho

Choline

Glx

Glutamate + glutemine

TR

Repetition time

TE

Echo time

MM

Macromolecules

SD

Standard deviation

HC

Healthy control

SN

Substantia nigra

GM

Grey matter

WM

White matter

MEGA-PRESS

Mescher–Garwood Point Resolved Spectroscopy

MPRAGE

Magnetization-prepared rapid-acquisition gradient echo

NMSQ

Non-Motor Symptom Questionnaire

SPSS

Statistical Package for Social Sciences software

UPDRS

Unified Parkinson’s Disease Rating Scale

H-Y stage

Hoehn and Yahr Stage

SB

Sleep bruxism

OSA

Obstructive sleep apnea