The study was approved by the Local Ethics Committee on Human Research of Kaohsiung Chang Gung Memorial Hospital in Taiwan. All the participants or their next of kin provided written informed consent prior to participation in the study. Thirty-one right-handed PD patients (17 men and 14 women, mean age: 60.39 ± 13.03 years) with no previous history of neurological or psychiatric illnesses, psychotropic medication, or contraindication to magnetic resonance imaging (MRI) were prospectively enrolled at the Neurology Department of Kaohsiung Chang Gung Memorial Hospital. Patients were included if they had idiopathic PD diagnosed by an experienced neurologist using the Parkinson’s Disease Society’s criteria for idiopathic PD (13). Of the 31 PD patients, 12 had never used any PD medication and 19 were taking dopaminergic medications (e.g., levodopa and dopamine agonists). The disease severity and functional status of each patient were evaluated on the Unified Parkinson’s Disease Rating Scale (UPDRS) (14), the modified Hoehn and Yahr scale (HY stage) (15), and the Schwab and England activities of daily living (SE-ADL) scale (16) in the “OFF” state, which refers to the period of lower levodopa effectiveness in eliminating motor symptoms and consequent inability to function properly.

The mean disease duration, defined as the time since symptoms first became subjectively apparent, was 4.03 ± 4.20 years. For comparison, 37 sex- and age-matched healthy subjects [24 men and 13 women, mean age: 55.49 ± 8.05 years; normal control (NC) group] with no medical history of neurologic diseases or psychiatric illnesses, alcohol/substance abuse, or head injury, and with similar levels of education, were also recruited.

The Mini-Mental State Examination (MMSE) (17) and Cognitive Ability Screening Instrument (CASI) (18) were administered to all the subjects. The CASI has a score range of 0–100 and provides a quantitative assessment of attention, concentration, orientation, short-term memory, long-term memory, language abilities, visual construction, list-generating fluency, abstraction, and judgment. The typical administration time for the CASI is 15–20 min.

The demographic data, including age, sex, and education, were compared among the study groups using a two-sample Student’s t-test and Pearson chi-square test, where appropriate, and are reported as the mean ± the SD. Differences in neuropsychological assessments were analyzed using an analysis of covariance model with the participant’s age, sex, and education values as covariates.

The demographic and clinical data of the participants are shown in Table 1. Sex, age, and education level were similar between the PD patients and the NC group. The PD patients had a mean total UPDRS score of 41.6, a mean modified HY stage of 2.2, and a mean SE-ADL score of 80.7.

The PD group revealed worse cognition, including lower MMSE (22.43 ± 6.20 versus, 27.59 ± 2.35; P < 0.001) and CASI scores (77.96 ± 21.31 versus 89.70 ± 8.19; P = 0.002) than the NC group (Table 1).

The striatal and subregional TRODAT-1 BP_ND results are presented in Table 1. The average TRODAT-1 BP_ND value was 1.57 ± 0.20 in the right caudate, 1.47 ± 0.22 in the right putamen, 2.02 ± 0.20 in the right striatum, 1.90 ± 0.18 in the left caudate, 2.09 ± 0.20 in the left putamen, and 2.01 ± 0.17 in the left striatum.

Our rs-fMRI analysis demonstrated that functional connections were segregated in the striatum and roughly parallel in the neocortices of both the NC group and the PD patients (Figure 1). Two-sample t-testing clearly showed significant between-group differences in the functional connectivity maps of each striatal seed (Figure 2) (P < 0.05, AlphaSim corrected). The hot and cold colors in the functional connectivity maps projected onto the surface of the normalized brain template indicated that the NC group had some functional correlation maps that were greater than and others that were smaller than those of the PD patients. Spatial maps and anatomical descriptions of each seed can be found in Table 2.

Decreased striatal and subregion TRODAT-1 BP_ND values were associated with changes in fc-cc in 4 regions of the bilateral Vsi functional network (Table 3; Figure 3). The lower fc-cc between the Vsi and right superior medial frontal (SMF) lobe was correlated with lower TRODAT-1 BP_ND values in the right striatum, right putamen, left striatum, left caudate, and left putamen. The lower fc-cc between the Vsi and left superior frontal lobe was correlated with lower TRODAT-1 BP_ND values in the left striatum and left caudate.

The higher fc-cc between the Vsi and right cerebellum was correlated with lower a TRODAT-1 BP_ND value in the left caudate. The higher fc-cc between the Vsi and left parahippocampus was correlated with a lower TRODAT-1 BP_ND value in the right caudate.

Among the aforementioned fc-cc values, the fc-cc values between the Vsi and right SMF lobe was correlated with lower TRODAT-1 BP_ND values in the bilateral striatum and bilateral putamen based on a threshold of P < 0.05 corrected for multiple comparisons.

The fc-cc difference between the Vsi and SMF lobe was also significantly correlated with increased disease duration (r = −0.533, P = 0.004) (Figure 4), higher HY stage (r = −0.431, P = 0.020), and lower SE-ADL scale score (r = 0.369, P = 0.049). Using a P value of 0.05 corrected for multiple comparisons, the fc-cc difference between the Vsi and SMF lobe was significantly correlated with increased disease duration. There was no correlation between altered striatal functional networks and the neuropsychological test results.

This resting-state functional connectivity study compared patients with idiopathic PD to age-matched control subjects. Using seed ROI manually defined bilaterally in six subdivisions of the striatum, we found: (i) lower striatal correlation with the extended frontal lobe, temporal lobe, rolandic operculum, angular gyrus, supramarginal gyrus, cingulate gyrus, striatum, insula, and cerebellum; (ii) increased striatal correlation with the cerebellum, supplementary motor cortex, parahippocampus, postcentral gyrus, thalamus, and occipital lobe; (iii) different effects on the putamen and caudate (Figure S1 in Supplementary Material), and most importantly (iv) correlation of disease severity with striatal dopamine neuron loss resulting from particular striatocortical network alterations. Our results support our previous hypothesis and provide clues to the pathophysiology of PD.

Using rs-fMRI analysis, first, we demonstrated a ventromedial-to-dorsolateral gradient in the striatum with a roughly parallel gradient in the neocortices of both the NC group and the PD patients (Figure 1) (12). The finding was consistent with our previous striatal structural covariance network findings of a correlation between the mean gray matter volume of the striatum with the gray matter volume across the entire neocortex in a voxel-wise manner (23). The high similarity in the distribution of subregional striatocortical networks between the PD and NC groups supported the hypothesis that human neural networks can be defined by synchronous neural activity, a common corticotrophic destiny, and selective vulnerability to neurodegenerative illness (24). It also supported the network degeneration hypothesis and suggested that pathological processes occurring in the striatum of PD brains affect the same striatocortical networks that co-vary in the striatum of healthy brains (24). Nevertheless, our patterns of rs-fMRI networks of the same striatal subdivisions in the NC and PD participants were wider than those mentioned in structural covariance studies conducted previously (25). The discrepancy between these results may be explained by the tendency of neuronal homeostasis to be maintained by functional plasticity instead of structural plasticity (26), unless the capacity of neurons deteriorates as a result of the disease and structural remodeling subsequently increases to compensate for functional impairment. Another explanation is that most information within striatocortical circuit models is indirectly conveyed through specific thalamic nuclei (27), which might also minimize the necessity of structural similarity between the striatum and cortex.

Compared to the NC group, the PD patients showed widespread striatocortical functional network alterations (Figure 2). The caudate, especially the Vsi subregion, showed a decrease in connectivity with the frontal lobe, temporal lobe, parietal lobe, and some cerebellar regions and an increase in connectivity with the parahippocampus, some cerebellar regions, and the occipital lobe. Plausibly, the topography of the affected regions reflects the phenotype of PD, which is primarily a movement disorder with associated executive deficits (28) and visual disturbances (29). Our findings also support the association of parkinsonism with recently described circuits connecting the cerebellum to the basal ganglia (10). Functional or morphological modulations in the cerebellum are related to akinesia/rigidity, tremor, gait disturbance, dyskinesia, and some non-motor symptoms (10). We also found that underactivation of the ventral attentional network in the supramarginal gyrus, anterior cingulate cortex, and insula may underlie the characteristic impairment of attentional mechanisms in PD (30).

Change in the striatocortical network was demonstrated, for the first time, to be linked to decline in the dopamine transporter ratio of the striatum, suggesting that change in striatal dopamine activity profoundly influences corticostriatal connectivity. We found that decreased dopamine activity was associated with functional alterations, mainly between the caudate seed and areas of the dopaminergic cortex, such as the frontal lobe, limbic system, and cerebellum (Figure 3). For example, dopamine depletion disrupts basal ganglia outflow and consequently affects the expression of prefrontal executive functioning by interrupting frontostriatal circuitry in PD (31). Our results were further strengthened by showing correlations between decreased numbers of striatofrontal network connections and increased disease severity and duration (Figure 4). Our results are open to several interpretations. Optimal dopamine levels are important, especially for neuroplasticity in the prefrontal cortices and the limbic system (32). Dopamine neuron degradation thus directly or indirectly disrupts both adjacent and distant areas of connection that share vulnerability to an altered neurotransmitter-related neuronal circuitry. In addition, sustained attenuation of dopamine activity weakens the protective effect of dopamine against glutamate-induced excitotoxicity. Abundant glutamate receptors within the frontal cortex, limbic system, and basal ganglia circuitry might account for cognitive and motor alterations in patients with PD (33).

Consistent with a previous structural covariance study (23), this functional network study revealed that the number of connections between the Vsi of the caudate and limbic system (including the cingulate gyrus, parahippocampus, insula, and temporal lobe) are increased in patients with PD. A recent PD longitudinal study also showed increased limbic system gray matter volume (34). Meanwhile, a report showed reduced dopamine transporters in the striatum and limbic system in PD patients with depression (35). Although our functional network results did not correlate with clinical presentation of the PD patients, this observation might also suggest compensation by the anterior striatum for the diseased posterior striatum.

We found increases in the striatocerebellar connectivity in some of the patients with PD while also finding decreases in such connectivity in others. Some of our findings were consistent with those of previous functional network and structural covariance studies (23, 36). In fact, both pathological and compensatory effects of the cerebellum have been observed in PD (10). Dopaminergic degeneration with subsequent structural damage and functional disconnection can account for the decrease in connectivity (10). Functional MRI studies showed weakened striatal–cortical and striatal–cerebellar connections causing basal ganglia dysfunction. The weakening of these connections can result, however, in compensatory strengthening of connections between corticocerebellar motor regions (36). In the present study, our finding of an association between decreased striatal dopamine levels and increased Vsi–cerebellar functional connectivity is consistent with the finding of cerebellar hyperactivity in PD patients by a previous motor task study (37). However, whether cerebellar hyperactivity is pathological or compensatory and how it affects phenotypes in PD patients remain open questions. We believe that a balance between different neurotransmitter alterations in PD might contribute to function modulation. Yet, this concept requires testing in further studies.

The results presented here should be interpreted with caution. It is possible, for example, that the observed group differences reflect long-term therapeutic effects rather than the effects of PD itself. Furthermore, the lack of any SPECT data for the NC group was another limitation of this study, one that is related to the reluctance of NCs to undergo SPECT scans. In any case, the lack of a direct comparison in this regard of the treated group to the untreated group may be considered a limitation. Some factors affecting image quality may have partially affected the study evaluation. Motion-related artifacts can affect rs-fMRI studies of PD patients in OFF medication status, and while the effects can be reduced using regression techniques, the effectiveness of this strategy is uncertain (38). Otherwise, our study failed to explore the relationships among dopamine neuron loss, putamen-associated networks, and clinical presentation. Since the HY stage or SE-ADL test was designed to evaluate general disease severity rather than specific functional domains in PD, our correlation results might reflect an evaluation bias for the frontal lobe, cerebellum, and other regions with functional network differences between the groups. In addition, the longer disease duration (mean 4.03 ± 4.20 years) in the patients with affected extensive extra-striatal GM makes it difficult to evaluate early putamen-associated motor network alterations in the present study. The study nonetheless partially reveals the natural course of PD under medical control and may be reinforced by further investigations using detailed clinical and neuropsychological examinations.