Parkinson’s disease (PD) is hypothesized to arise from a complex interplay between genetic and environmental factors; however, its precise etiology remains unclear. The most widely accepted pathological hallmark of PD is dopaminergic neurons degeneration in the substantia nigra pars compacta, which disrupts striatal dopamine regulation, thus ultimately leading to motor dysfunction (Kalia and Lang, 2015). These neurons are closely associated with the cerebral microvasculature; their degeneration reduces metabolic demand, thereby altering regional cerebral perfusion patterns (Kim et al., 2016). Recent neuroimaging studies have identified blood–brain barrier (BBB) disruption as a critical pathogenic feature in neurodegenerative mechanisms (Sweeney et al., 2018). In patients with PD, evidence of BBB dysfunction includes erythrocyte extravasation, protein accumulation within the striatum (Gray and Woulfe, 2015), thinning of the capillary endothelial layer, and widespread extraluminal immunoglobulin G (IgG) staining in the subthalamic nucleus (Pienaar et al., 2015). These findings suggest that PD might also represent a cerebrovascular disorder characterized by microvascular dysregulation (Nanhoe-Mahabier et al., 2009; Rane et al., 2020).

Arterial spin labeling (ASL), which is a noncontrast magnetic resonance imaging (MRI) modality for perfusion measurement, utilizes magnetically labeled blood water as an endogenous tracer. This technique quantifies cerebral blood flow (CBF) by applying radiofrequency pulses to magnetically invert the arterial blood proximal to the imaging region. However, conventional ASL methods have limitations to hemodynamic quantification owing to inter-individual variability and reliance on a single post-labeling delay (PLD), which may inaccurately represent heterogeneous blood flow velocities. To overcome these limitations, Cohen et al. (2020) pioneered an advanced multi-PLD ASL strategy termed multi-delay pseudo-continuous ASL (MD-pCASL), which enhances CBF accuracy by accounting for diverse arterial transit times (ATTs). This technique employs a short-burst radiofrequency pulse train to achieve high temporal resolution while maintaining an inversion efficiency comparable to that of continuous ASL. Additionally, MD-pCASL minimizes magnetization transfer effects and improves labeling efficiency, thereby yielding superior signal-to-noise ratios and broadening the clinical applicability of ASL. MD-pCASL quantifies two critical parameters: the ATT, which measures the duration of blood travel from the labeling plane to the imaging voxel, and ATT-corrected CBF, which provides a more precise estimate of capillary-level blood delivery rates by compensating for delayed arterial arrival.

Previous MRI studies have extensively documented neurovascular alterations in patients with PD. Wang et al. compared the diagnostic performance of conventional single-delay ASL and multi-delay ASL in PD cases and demonstrated that integrated models incorporating hemodynamic features from multi-delay ASL exhibited superior diagnostic utility. Compared with single-delay ASL, this approach offers detailed visualization of abnormal CBF and additional insights into ATT and cerebral blood volume (CBV), particularly in motor-associated regions (Wang et al., 2024). Similarly, Al-Bachari et al. (2017) found that patients with idiopathic PD exhibit diffuse ATT prolongation and focal CBF reduction. Notably, they were the first to report more widespread ATT prolongation in the tremor-dominant (TD) subtype than in the postural instability/gait difficulty subtype, despite observing no significant CBF differences between them.

Recent studies have validated MD-pCASL as a superior technique for assessing cerebrovascular disease, particularly for perfusion quantification (Yu et al., 2021; Luijten et al., 2023). However, the hemodynamic differences underlying the phenotypic divergence between the TD and akinetic–rigid-dominant (ARD) subtypes of PD remain poorly characterized. Elucidating these subtype-specific perfusion patterns might provide novel mechanistic insights into the clinical heterogeneity of PD, thereby bridging the gap between neurovascular pathophysiology and motor symptom manifestations. The current study aimed to investigate the differences in global cerebral hemodynamics between TD and ARD subtypes using MD-pCASL imaging and evaluate the diagnostic efficacy of multi-parameter MD-pCASL technique for motor subtypes.

This study employed the novel MD-pCASL technique to comprehensively assess CBF, ATT, and CBV values in the TD, ARD, and HC groups without contrast administration. Our findings revealed that the ARD group exhibited elevated CBF and CBV in multiple left-hemispheric cortical regions compared with the HC group, whereas no hyperperfused areas were identified in the TD group. Furthermore, direct comparisons between the ARD and TD subgroups showed significantly higher perfusion levels in the ARD group, particularly in the left hippocampal formation and basal ganglia nuclei. Notably, our results showed marked hemispheric lateralization, with the most significant differences localized in the left cerebral hemisphere, a phenomenon potentially linked to predominantly right-sided motor symptom onset. This observation aligns with the seminal work of Shang et al. (2021), which posited that lateralized motor manifestations in patients with PD might reflect asymmetrical regional and interregional CBF abnormalities, thereby underscoring the necessity for stratifying PD subtypes based on motor laterality patterns.

Wang et al. compared the hemodynamic abnormalities in the brain of patients with PD on multiple delayed arterial spin labeling and the traditional single delayed arterial spin labeling technique. They found that the corrected CBF demonstrated better classification performance than did the uncorrected CBF. Moreover, the AUC of both CBV and ATT was relatively high. The comprehensive model emphasizing the hemodynamic characteristics of MDASL provides better performance for PD diagnosis (Wang et al., 2024). However, this study did not conduct a classification discussion about PD and ignored the influence of subtype heterogeneity. Based on this, our study demonstrated the different hemodynamic characteristics among the motor subtypes. The elevated CBF observed in the ARD subgroup suggests localized BBB hyperpermeability and greater neurovascular unit disruption (Li et al., 2020). Hippocampal formation—a hub for memory consolidation, spatial navigation, and emotional modulation—exhibited hyperperfusion in ARD cases, thus potentially contributing to its accelerated cognitive decline compared with the TD subtype (Wickremaratchi et al., 2011). Postmortem analyses of brains from patients with PD have revealed perivascular fibrinogen/fibrin (Gray and Woulfe, 2015), IgG (Pienaar et al., 2015), and hemosiderin (Loeffler et al., 1995) deposits within the striatal vasculature, which are indicative of chronic BBB breakdown. Al-Bachari et al. (2017) reported that the prolongation of ATT in the TD group was more diffuse. However, no differences in CBF among subtypes were found. Our research results have been supplemented in terms of CBF/CBV. Al-Bachari et al. (2017) also reported basal ganglia hyperperfusion and hypothesized that the exacerbated nigrostriatal pathway dysfunction in patients with ARD may drive this phenomenon, in contrast to the compensatory downstream neuromodulatory adaptations in those with TD (Zaidel et al., 2009). Notably, subthalamic nucleus deep brain stimulation–mediated tremor amelioration may involve microvascular remodeling (Sweeney et al., 2018), further implicating neurovascular crosstalk in PD symptomatology.

Pharmacologically, levodopa, which is the first-line therapy for PD, may influence basal ganglia perfusion. At therapeutic doses, levodopa is converted to dopamine by preserved dopaminergic neurons in the putamen, followed by a dysregulated release due to deficient autoreceptor feedback (Politis et al., 2014). The vasodilatory properties of dopamine enhance the regional CBF (Iadecola, 1998), potentially facilitating BBB-penetrant drug delivery. Pelizzari et al. (2019) identified positive correlations between UPDRS-III scores and CBF in the sensorimotor basal ganglia networks, a finding corroborated and extended by our study. Specifically, ARD exhibited stronger correlations with deep basal ganglia nuclear perfusion and motor severity, thus confirming the pivotal role of striatopallidal degeneration in ARD pathogenesis and progression.

This study did not identify significant differences in ATT between PD subgroups (ARD/TD) and HC. However, the ARD subgroup exhibited significantly prolonged ATT in the left middle temporal gyrus compared with the TD subgroup. Notably, studies on acute stroke have proposed that ATT prolongation may reflect preserved perfusion (Wolf et al., 2014) or collateral compensation (Zaharchuk, 2011), both mechanisms that warrant further investigation in neurodegenerative contexts. In addition, regression analysis and ROC curve analysis in this study showed that the combined indicators of CBF, CBV, and ATT had the best diagnostic efficacy in identifying PD subtypes. This preliminarily indicates that multi-parameter blood flow indicators can be used as effective imaging biological markers for distinguishing between motor subtypes.

P-glycoprotein, which is a critical cellular defense mechanism against neurotoxicant accumulation in the brain (Zhao et al., 2015), has been shown via positron emission tomography (PET) imaging to exhibit reduced activity in the midbrain of patients with PD, which is indicative of BBB dysfunction (Kortekaas et al., 2005). Previous quantitative PET studies have documented widespread cortical hypoperfusion and hypometabolism in the frontal–parietal–occipital regions of PD cohorts (Borghammer, 2012). Our findings further demonstrated that the ARD subgroup exhibited decreased CBF and CBV in the left angular gyrus and inferior parietal lobule compared with the HC group, with these hemodynamic reductions negatively correlating with MoCA and MMSE scores; this pattern is exacerbated by comorbid vascular pathologies and risk factors that are known to accelerate motor and cognitive decline in PD (Malek et al., 2016). Notably, diminished cerebral perfusion has also been reported in patients with mild cognitive impairment and often precedes structural atrophy (Kisler et al., 2017).

By contrast, the TD group showed reduced CBF and CBV in the left precuneus and cuneus, which are regions that are positively associated with olfactory function scores. Olfactory dysfunction, a hallmark prodromal feature of neurodegenerative diseases that precedes cognitive deficits (Murphy, 2019; Slabik and Garaschuk, 2023), persists throughout the disease course of TD-PD and may even antedate motor symptom onset (Wang et al., 2022). Although the cuneus and precuneus primarily contribute to visuospatial processing, their involvement in olfactory function may arise from visuo-olfactory integration or indirect modulation via default mode network pathways. Huang et al. (2024) investigated the neuroimaging mechanisms underlying olfactory regulation in patients with cerebral small vessel disease and similarly identified reduced regional homogeneity and functional connectivity in the left cuneus in those with PD, which correlated positively with olfactory performance.

This study has a few limitations that warrant consideration. First, the modest sample size necessitates the validation of statistical effects through larger multicenter cohorts in future investigations. Second, although MD-pCASL effectively mapped perfusion heterogeneity between TD and ARD groups, its mechanistic links to underlying neuropathological substrates—such as α-synuclein propagation or neurovascular unit degeneration—require further exploration in animal models or hemodynamic phantom studies. Third, our classification based on motor dysfunction phenotypes might have limited detection of nonmotor symptoms associated with cerebral perfusion, thus underscoring the need for transdiagnostic biomarkers in future PD subtyping frameworks. Fourth, the lack of standardized medication washout protocols (e.g., levodopa or dopamine agonists) might have obscured symptom severity or perfusion responses, a confounder that should be adjusted for in future longitudinal designs.

In conclusion, the MD-pCASL technology accurately identifies the specific change patterns of cerebral blood perfusion of TD and ARD types in patients with PD by optimizing multi-parameter hemodynamic detection. At the same time, it also confirms the clinical application potential of the MD-pCASL technology in the early differential diagnosis, subtype classification and disease course monitoring for PD. It is expected to promote the practical application of precision medicine in the field of neurodegenerative diseases.