Dopaminergic‐ and Cholinergic‐Related Structural MRI Patterns Across Cognitive Subtypes of Parkinson's Disease With Mild Cognitive Impairment Authors: Xueqin Bai, Haoxin Chen, Qiuyue Li, Shiwei Zhang, Shuangli Chen, Andan Qian, Ronghui Zhou, Yitong Cheng, Chenglong Xie, Zhen Wang, Yunjun Yang, Meihao Wang Published: 2026 ABSTRACT ABSTRACTBackgroundThe dual syndrome hypothesis distinguishes frontostriatal and posterior cortical cognitive subtypes in Parkinson's disease with mild cognitive impairment (PD‐MCI), which are associated with distinct cognitive profiles and different risks of dementia.ObjectiveTo characterize dopaminergic‐ and cholinergic‐related structural MRI patterns across cognitive subtypes of PD‐MCI.MethodsA total of 182 non‐demented patients with Parkinson's disease (PD) and 38 healthy controls (HC) were included. Based on comprehensive neuropsychological assessments, PD patients were classified as having normal cognition (PD‐NC, n = 112), PD‐MCI with frontostriatal deficits (n = 20), posterior cortical deficits (n = 28), or mixed deficits (n = 22). All participants underwent T1‐weighted MRI. Normalized volumetric measures of subcortical nuclei (caudate, putamen, globus pallidus, and thalamus) and the basal forebrain were obtained.ResultsDistinct subtype‐specific patterns of subcortical and basal forebrain atrophy were observed. PD‐MCI patients with posterior cortical deficits showed significant atrophy of cholinergic nucleus 4 (Ch4) compared with HC and PD‐MCI patients with frontostriatal deficits. In contrast, PD‐MCI patients with frontostriatal deficits exhibited atrophy of the caudate nucleus and putamen relative to HC. The mixed deficits subtype showed putamen and thalamic atrophy compared with HC. In addition, basal forebrain Ch4 volume was significantly associated with visuospatial performance in PD‐MCI patients.ConclusionsDifferent PD‐MCI subtypes exhibit distinct patterns of dopaminergic‐ and cholinergic‐related structural alterations on MRI. Posterior cortical deficits are predominantly associated with cholinergic basal forebrain degeneration, whereas frontostriatal deficits are linked to dopaminergic involvement. These findings support the neural heterogeneity underlying cognitive impairment in PD. Introduction Parkinson's disease (PD) is the second most common neurodegenerative disorder, and cognitive impairment is one of the most prevalent and debilitating non‐motor symptoms in PD. Previous studies have reported that approximately 40% of PD patients experience cognitive impairment [1]. PD with mild cognitive impairment (PD‐MCI) represents individuals with an early stage of cognitive impairment and is considered a major risk factor for the subsequent development of dementia. Investigating PD‐MCI may provide important insights into the early neurobiological mechanisms underlying cognitive impairment in PD. Notably, the clinical profile of PD‐MCI is highly heterogeneous, involving impairments across multiple cognitive domains [2]. Moreover, the clinical course of PD‐MCI is variable, with some patients progressing rapidly to dementia, whereas others remain cognitively stable or even revert to normal cognition [3, 4]. Thus, elucidating the neural mechanisms underlying this heterogeneity is crucial for improving prognostic accuracy and facilitating the development of targeted therapeutic strategies. The core pathological hallmark of PD is the deposition of α‐synuclein in the brain. Neuropathological studies have demonstrated that the abnormal accumulation of α‐synuclein contributes to widespread neurodegeneration, affecting multiple neurotransmitter systems [5, 6]. Traditionally, PD has been primarily attributed to dysfunction of the dopaminergic frontostriatal circuitry. Nevertheless, dopaminergic depletion alone cannot fully account for the complexity of cognitive decline in PD, as dopaminergic medications show limited efficacy in non‐executive domains such as visuospatial, memory, and language functions [7]. Growing evidence suggests that degeneration of the cholinergic system, particularly within the basal forebrain cholinergic nuclei, plays a critical role in the pathogenesis and progression of cognitive impairment in PD patients [8, 9]. Accordingly, cognitive impairment in PD patients is likely influenced by abnormalities in both the dopaminergic and cholinergic systems, and an integrated analysis of dopaminergic‐ and cholinergic‐related degeneration may provide deeper insight into the heterogeneous clinical manifestations of PD‐MCI. The dual syndrome hypothesis proposes two cognitive subtypes of PD‐MCI: The frontostriatal subtype characterized by executive and attentional dysfunction, and the posterior cortical subtype characterized by visuospatial and memory dysfunction [10]. Notably, the posterior cortical subtype has been associated with a higher risk of progression to dementia. Previous studies have demonstrated cortical and subcortical atrophy in PD‐MCI and Parkinson's disease with dementia (PDD) by using structural magnetic resonance imaging (MRI), suggesting that neurodegeneration contributes substantially to cognitive impairment in PD. However, most of these studies have treated PD‐MCI or PDD as a single clinical entity [11, 12, 13], despite its well‐recognized clinical heterogeneity. The structural basis underlying different subtypes of PD‐MCI remains incompletely understood. More recently, neuroimaging studies have reported distinct patterns of brain activation and functional connectivity across different PD‐MCI subtypes [14, 15, 16]. These findings support the concept that different cognitive phenotypes in PD‐MCI may arise from distinct neural mechanisms. Nevertheless, most previous investigations have been conducted using whole‐brain approaches and have not specifically examined degeneration within neurotransmitter‐related systems. Given the critical roles of both dopaminergic and cholinergic dysfunction in cognitive impairment of PD, it remains unclear whether different PD‐MCI subtypes exhibit distinct dopaminergic‐ and cholinergic‐related structural degenerative patterns. In the present study, we employed structural MRI to quantify volumetric alterations in brain regions associated with the dopaminergic and cholinergic systems, including the basal forebrain and subcortical nuclei (caudate, putamen, globus pallidus, and thalamus). By simultaneously characterizing dopaminergic‐ and cholinergic‐related structural alterations across distinct PD‐MCI subtypes, we aimed to identify subtype‐specific neurodegenerative patterns and provide imaging evidence to elucidate the neurobiological underpinnings of cognitive heterogeneity in PD. Methods Subjects All data in this study were obtained from the Parkinson's Progression Markers Initiative (PPMI) database downloaded on December 1, 2024. PPMI is an ongoing prospective, multi‐center study designed to identify clinical, imaging, genetic, and biospecimen biomarkers of PD progression. Study protocols and manuals are available on the PPMI website (https://www.ppmi‐info.org). The PPMI study protocol was approved by the institutional review boards of all participating sites, and all participants signed informed consent prior to enrollment. PD participants enrolled in PPMI were required to (1) have a recent idiopathic PD diagnosis, (2) be untreated for PD, (3) demonstrate a dopamine transporter (DAT) deficit. PD and healthy control (HC) participants who underwent a comprehensive cognitive assessment, including global cognitive function tests and a neuropsychological battery covering all five cognitive domains (with two tests per domain) at baseline, were included in this study. The exclusion criteria were as follows: (1) PD participants with dementia; (2) unavailable or poor‐quality T1‐weighted MRI data. The participant flowchart is shown in Figure 1. Flowchart shows participants' enrollment. Abbreviations: HC, healthy controls; PD, Parkinson's disease; PD‐NC, Parkinson's disease with normal cognition; PD‐FS, Parkinson's disease with mild cognitive impairment‐frontostriatal deficits subtype; PD‐PC, Parkinson's disease with mild cognitive impairment‐posterior cortical deficits subtype; PD‐MS, Parkinson's disease with mild cognitive impairment‐mixed subtype. All participants received the standard PPMI assessment battery. Demographic and clinical information, including age, sex, and years of education, were obtained at baseline. The motor characteristics of PD patients were assessed using the Movement Disorder Society Unified Parkinson's Disease Rating Scale (MDS‐UPDRS) (Parts I through III) and the Hoehn and Yahr (HY) stage. Depressive symptoms were assessed using the Geriatric Depression Scale (GDS); rapid eye movement sleep behavior disorder was evaluated with the Rapid Eye Movement Sleep Behavior Disorder Questionnaire (REM); and autonomic symptoms were assessed using the Scales for Outcomes in Parkinson's Disease–Autonomic (SCOPA). All the subjects underwent a comprehensive cognitive assessment including global cognitive function tests and a neuropsychological battery covering all five cognitive domains. Global cognitive function was assessed using the Montreal Cognitive Assessment (MoCA). The detailed neuropsychological battery encompassed five domains: (1) executive function, assessed by the Trail Making Test part B (TMT‐B) and the Symbol Digit Modalities Test (SDMT); (2) attention and working memory, assessed by the Trail Making Test part A (TMT‐A) and letter‐number sequencing (LNS); (3) visuospatial function, assessed using the Clock‐Drawing Test (CDT) and the Benton Judgment of Line Orientation Test (BJLOT); (4) language, assessed using the Boston Naming Test (BNT) and the Semantic Verbal Fluency Test (SFT); and (5) memory, evaluated using the Hopkins Verbal Learning Test (HVLT) with immediate recall and delayed recall. PD‐MCI was diagnosed according to the Level II diagnostic criteria recommended by the Movement Disorder Society (MDS) Task Force 2012 [2], defined as impairment on at least two tests for the detailed cognitive battery with scores 1.5 standard deviations (SDs) below the normative mean. PD patients without dementia who did not meet the criteria for PD‐MCI were classified as PD with normal cognition (PD‐NC). Based on the dual syndrome hypothesis, the PD‐MCI patients were further categorized into three subtypes: (1) frontostriatal subtype (PD‐FS): Deficits in attention/working memory and/or executive functions without deficits in visuospatial functions, episodic memory, and language; (2) posterior cortical subtype (PD‐PC): Deficits in visuospatial functions and/or episodic memory and/or language without deficits in attention/working memory and executive functions; and (3) mixed subtype (PD‐MS): Combined impairments in attention/working memory and/or executive functions together with deficits in visuospatial functions, episodic memory, and/or language [14]. MRI Data Acquisition Neuroimaging data were collected using 3.0 Tesla scanners across multiple PPMI sites following standardized acquisition protocols. For each MRI session, three‐dimensional (3D) T1‐weighted structural images were acquired using a magnetization‐prepared rapid gradient echo or spoiled gradient‐recalled echo sequence. The details of the data acquisition parameters could be found on the PPMI website (http://www.ppmiinfo.org/study‐design/research‐documents‐and‐sops/). MRI Data Post‐Processing The regions of interest (ROIs) of the dopaminergic and cholinergic systems were selected based on prior anatomical and neurobiological knowledge of dopaminergic and cholinergic systems. Specifically, the caudate, putamen, globus pallidus, and thalamus were selected as key nodes within the nigrostriatal dopaminergic circuitry [17], whereas the basal forebrain regions, including cholinergic nuclei 1–3 (Ch123) and cholinergic nucleus 4 (Ch4), were selected to represent the cholinergic system because they contain the major cholinergic nuclei projecting to the cerebral cortex [18]. A schematic illustration of the ROIs is provided in the Supporting Information (Supplementary Figures S1 and S2) for improved anatomical clarity. Basal forebrain volumes were quantified using an atlas‐based approach. The 3D T1‐weighted MRI data were processed using the Computational Anatomy Toolbox 12 within Statistical Parametric Mapping (SPM12, https://www.fil.ion.ucl.ac.uk/spm/) for segmentation, spatial normalization, and volumetric analysis. The ROIs of the basal forebrain were defined according to the probabilistic anatomical maps implemented in the SPM Anatomy Toolbox, derived from the 3D reconstruction of histological sections from 10 postmortem human brains [19]. Based on cytoarchitecture, the basal forebrain was separated into Ch123 and Ch4. The basal forebrain volumes were then automatically extracted by summing the modulated gray matter voxel values within the corresponding ROIs in MNI standard space [20]. The volumes of basal ganglia nuclei were acquired by the FreeSurfer software (version 6.0.0, http://surfer.nmr.mgh.harvard.edu/fswiki). The 3D T1‐weighted images underwent automated preprocessing via the “recon‐all” pipeline, including motion correction, bias field correction, skull stripping, and tissue segmentation, generating labeled volumes (aparc+aseg.mgz) with subcortical structures segmented according to FreeSurfer's atlas [21]. The ROIs of basal ganglia nuclei include bilateral caudate, putamen, globus pallidus, and thalamus. The values from bilateral ROIs were averaged. Total intracranial volume (TIV) was also acquired for each participant. Statistical Methods Demographic and clinical data were compared among groups using one‐way analysis of variance (ANOVA) followed by post hoc Bonferroni correction for multiple comparisons, or Kruskal–Wallis test followed by Bonferroni method for multiple‐comparison correction. Differences in sex distribution were assessed using a chi‐square test. Group differences in basal ganglia nuclei and basal forebrain volumes among groups were examined using analysis of covariance (ANCOVA), with age, sex, education years, and TIV included as covariates. Partial correlation analyses were then performed to investigate associations between regional brain volumes and cognitive performance in PD patients, controlling for the same covariates. To address multiple comparisons, results were adjusted using the False Discovery Rate (FDR) correction. Statistical significance was set to two‐tailed p < 0.05. Results The demographic and clinical characteristics of all subjects are summarized in Table 1. According to the detailed neuropsychological assessments, 112 PD patients were classified as PD‐NC and 70 patients were classified as PD‐MCI. Within PD‐MCI patients, 20 (28.6%) were categorized as the frontostriatal subtype (PD‐FS), 28 (40.0%) as the posterior cortical subtype (PD‐PC), and 22 (31.4%) as the mixed subtype (PD‐MS). TABLE 1Demographic and clinical feature comparisons among groups.HC (n = 38)PD‐NC (n = 112)PD‐FS (n = 20)PD‐PC (n = 28)PD‐MS (n = 22) p Post hoc test Age(years) 65.26 ± 7.5063.21 ± 10.5464.11 ± 7.3161.07 ± 9.5763.57 ± 9.090.571NA Sex(male/female) 18/2064/4813/718/1014/180.340NA Education(years) 17.00 ± 1.9616.43 ± 2.3616.07 ± 3.4915.46 ± 2.6014.71 ± 2.690.001 HC>PD‐MS; PD‐NC>PD‐MS H‐Y stage —2 (1–2)2 (1–2)2 (1–2)2 (1–2)0.286NA MDS‐UPDRS I —7.42 ± 4.706.14 ± 3.666.71 ± 4.968.45 ± 5.700.406NA MDS‐UPDRS II —6.68 ± 4.487.07 ± 3.036.08 ± 5.407.58 ± 5.710.669NA MDS‐UPDRS III —21.55 ± 9.4720.23 ± 9.1120.70 ± 9.0226.17 ± 9.560.071NA GDS 1.98 ± 3.181.94 ± 2.123.07 ± 3.582.83 ± 3.932.42 ± 2.430.302NA REM 2.24 ± 2.534.28 ± 3.034.64 ± 3.184.30 ± 3.553.65 ± 2.410.001HCPD‐MS; PD‐NC>PD‐MS Abbreviations: HC, healthy controls; PD‐NC, Parkinson's disease with normal cognition (PD‐NC); PD‐FS, Parkinson's disease with mild cognitive impairment‐frontostriatal deficits subtype; PD‐PC, Parkinson's disease with mild cognitive impairment‐posterior cortical deficits subtype; PD‐MS, Parkinson's disease with mild cognitive impairment‐mixed subtype; H‐Y, Hoehn and Yahr; MDS‐UPDRS, Movement Disorder Society Unified Parkinson's Disease Rating Scale; GDS, Geriatric Depression Scale; REM, Rapid eye movement; SCOPA, Scales for Outcomes in Parkinson's Disease–Autonomic; MoCA, Montreal Cognitive Assessment. Demographic and clinical feature comparisons among groups. HC>PD‐MS; PD‐NC>PD‐MS HC>PD‐MS; PD‐NC>PD‐MS Abbreviations: HC, healthy controls; PD‐NC, Parkinson's disease with normal cognition (PD‐NC); PD‐FS, Parkinson's disease with mild cognitive impairment‐frontostriatal deficits subtype; PD‐PC, Parkinson's disease with mild cognitive impairment‐posterior cortical deficits subtype; PD‐MS, Parkinson's disease with mild cognitive impairment‐mixed subtype; H‐Y, Hoehn and Yahr; MDS‐UPDRS, Movement Disorder Society Unified Parkinson's Disease Rating Scale; GDS, Geriatric Depression Scale; REM, Rapid eye movement; SCOPA, Scales for Outcomes in Parkinson's Disease–Autonomic; MoCA, Montreal Cognitive Assessment. There was no significant difference in age or sex distribution among all groups. Years of education were higher in the HC and PD‐NC groups compared with the PD‐MS group. Among PD patients, no significant differences were found in H‐Y stage, MDS‐UPDRS part I, II, or III scores. There was no significant difference in depressive symptom scores among all groups, whereas the PD‐NC group showed higher REM and SCOPA scores compared with HC. Compared with both HC and PD‐NC, the PD‐MS group showed significantly lower MoCA scores. For domain‐specific cognitive performance, significant group differences were observed across all neuropsychological measures except for the Clock Drawing Test (Table 2). TABLE 2Neuropsychological battery of the subjects.HC (n = 38)PD‐NC (n = 112)PD‐FS (n = 20)PD‐PC (n = 28)PD‐MS (n = 22) p Post hoc test Executive function TMT‐B78.65 ± 38.7272.72 ± 27.15136.14 ± 62.8262.71 ± 12.27134.35 ± 58.84< 0.001 HC, PD‐NC, PD‐PCPD‐MS HC, PD‐NC, PD‐PC>PD‐FS Attention and working memory function TMT‐A35.04 ± 15.6931.40 ± 10.3056.36 ± 21.6029.75 ± 7.4147.10 ± 18.59< 0.001 HC, PD‐NC, PD‐PC>PD‐MS HC, PD‐NC, PD‐PC>PD‐FS LNS10.98 ± 2.9011.21 ± 2.528.36 ± 3.1011.58 ± 1.958.77 ± 2.32< 0.001 HC, PD‐NC, PD‐PC>PD‐MS HC, PD‐NC, PD‐PC>PD‐FS Visuospatial function BJLOT12.80 ± 2.3113.11 ± 1.8713.50 ± 1.3511.50 ± 1.8211.48 ± 2.05< 0.001 HC, PD‐NC, PD‐PC>PD‐MS PD‐NC, PD‐FS>PD‐PC CDT6.11 ± 1.436.50 ± 1.066.07 ± 1.06.42 ± 1.126.42 ± 1.110.115NA Language function BNT54.89 ± 8.8057.17 ± 3.8656.36 ± 4.9456.92 ± 2.7852.13 ± 9.95< 0.001PD‐NC, PD‐PC>PD‐MSSFT22.28 ± 5.5122.34 ± 4.9221.17 ± 5.6117.64 ± 4.6718.06 ± 5.56< 0.001 HC, PD‐NC>PD‐PC HC, PD‐NC>PD‐MS Memory function HVLT ImmediateRecall 25.50 ± 5.3525.72 ± 4.3323.21 ± 4.5422.46 ± 4.4418.87 ± 4.95< 0.001 HC, PD‐NC, PD‐FS, PD‐PC>PD‐MS; PD‐NC>PD‐PC HVLT Delayed Recall 9.02 ± 2.769.38 ± 1.958.00 ± 1.964.404 ± 2.855.16 ± 2.89< 0.001 HC, PD‐NC, PD‐FS>PD‐PC HC, PD‐NC, PD‐FS>PD‐MS Abbreviations: HC, healthy controls; PD‐NC, Parkinson's disease with normal cognition (PD‐NC); PD‐FS, Parkinson's disease with mild cognitive impairment‐frontostriatal deficits subtype; PD‐PC, Parkinson's disease with mild cognitive impairment‐posterior cortical deficits subtype; PD‐MS, Parkinson's disease with mild cognitive impairment‐mixed subtype. TMT‐B, Trail Making Test part B; SDMT, Symbol Digit Modality Test; TMT‐A, Trail Making Test part A; LNS, letter‐number sequencing; BJLOT, Benton Judgment of Line Orientation Test; CDT, Clock‐Drawing Test; BNT, Boston Naming Test; SFT, Semantic Verbal Fluency Test; HVLT, Hopkins Verbal Learning Test. Neuropsychological battery of the subjects. HC, PD‐NC, PD‐PCPD‐MS HC, PD‐NC, PD‐PC>PD‐FS HC, PD‐NC, PD‐PC>PD‐MS HC, PD‐NC, PD‐PC>PD‐FS HC, PD‐NC, PD‐PC>PD‐MS HC, PD‐NC, PD‐PC>PD‐FS HC, PD‐NC, PD‐PC>PD‐MS PD‐NC, PD‐FS>PD‐PC HC, PD‐NC>PD‐PC HC, PD‐NC>PD‐MS HVLT ImmediateRecall HC, PD‐NC, PD‐FS, PD‐PC>PD‐MS; PD‐NC>PD‐PC HVLT Delayed Recall HC, PD‐NC, PD‐FS>PD‐PC HC, PD‐NC, PD‐FS>PD‐MS Abbreviations: HC, healthy controls; PD‐NC, Parkinson's disease with normal cognition (PD‐NC); PD‐FS, Parkinson's disease with mild cognitive impairment‐frontostriatal deficits subtype; PD‐PC, Parkinson's disease with mild cognitive impairment‐posterior cortical deficits subtype; PD‐MS, Parkinson's disease with mild cognitive impairment‐mixed subtype. TMT‐B, Trail Making Test part B; SDMT, Symbol Digit Modality Test; TMT‐A, Trail Making Test part A; LNS, letter‐number sequencing; BJLOT, Benton Judgment of Line Orientation Test; CDT, Clock‐Drawing Test; BNT, Boston Naming Test; SFT, Semantic Verbal Fluency Test; HVLT, Hopkins Verbal Learning Test. With respect to basal forebrain volumes, both PD‐PC and PD‐MS exhibited significant atrophy of the cholinergic Ch4 compared with HC (p = 0.006 and p = 0.010, respectively). Furthermore, relative to PD‐FS patients, PD‐PC showed Ch4 atrophy (p = 0.025) (Figure 2‐A). Partial correlation analyses showed that the volume of Ch4 was significantly correlated with visuospatial performance as assessed by the BJLOT score in PD‐MCI patients (r = 0.33, p = 0.005) (Figure 3). Ch4 and subcortical volumes across PD‐MCI subtypes and healthy controls. (A) Ch4 volume comparisons among groups. (B) Caudate volume comparisons among groups. (C) Putamen volume comparisons among groups. (E) Thalamus volume comparisons among groups. Abbreviations: HC, healthy controls; PD‐NC: Parkinson's disease with normal cognition (PD‐NC); PD‐FS: Parkinson's disease with mild cognitive impairment‐frontostriatal deficits subtype; PD‐PC: Parkinson's disease with mild cognitive impairment‐posterior cortical deficits subtype; PD‐MS: Parkinson's disease with mild cognitive impairment‐mixed subtype. * indicates p < 0.05 and ** indicates p < 0.01. Association between Ch4 volume and visuospatial performance in PD‐MCI. Abbreviation: BJLOT, Benton Judgment of Line Orientation Test. Regarding subcortical volumes, compared with HC, the PD‐FS group exhibited significant atrophy of the caudate nucleus (p = 0.021) (Figure 2‐B) and putamen (p = 0.013) (Figure 2‐C). In addition, compared with HC, the PD‐MS group showed significant atrophy of the putamen and thalamus (p = 0.002) (Figure 2‐D). Discussion In this study, we investigated the dopaminergic‐ and cholinergic‐associated structural alterations across different phenotypes of PD‐MCI. Our findings demonstrate distinct patterns of basal ganglia and basal forebrain atrophy among PD‐MCI subtypes, highlighting the heterogeneity of neural substrates underlying cognitive impairment in PD. Specifically, compared with HC, the posterior cortical subtype (PD‐PC) and mixed subtype (PD‐MS) exhibited atrophy of the cholinergic nucleus Ch4. Notably, PD‐PC patients exhibited more pronounced Ch4 atrophy than the fronto‐striatal subtype (PD‐FS). In contrast, caudate nucleus atrophy was observed exclusively in the PD‐FS subtype compared with HC. In addition, PD‐MS exhibited significant atrophy of the putamen and thalamus compared with HC. Furthermore, across PD‐MCI patients, Ch4 volume was significantly associated with visuospatial performance, supporting the contribution of basal forebrain degeneration to cognitive deficits in PD. Our results revealed clear subtype‐specific differences in basal forebrain and basal ganglia involvement. Compared to HC, PD‐PC patients exhibited atrophy of Ch4. Moreover, Ch4 volume was significantly lower in PD‐PC than in the fronto‐striatal subtype (PD‐FS). Volume loss within the cholinergic basal forebrain has been shown to correlate with reduced neuronal density [22] and diminished cortical cholinergic synaptic activity [9]. Accordingly, the observed Ch4 atrophy in PD‐PC patients likely reflects prominent cholinergic degeneration. The cholinergic system exhibits widespread projections across the entire brain, with distinct basal forebrain nuclei innervating different cortical territories [18]. Specifically, Ch1–2 primarily project to the hippocampus and medial temporal lobe [23], whereas Ch4 provides extensive cholinergic input to widespread cortical regions [24], with the occipital and parietal cortex being more vulnerable. This topographic organization may help explain the present finding that atrophy was detected in Ch4 but not in Ch123 in PD‐PC patients. Besides, the volume of Ch4 was associated with visuospatial function in PD‐MCI patients in this study. Previous studies reported that cholinergic neurotransmitter binding was predominantly reduced in the occipital and parietal cortex in PD patients, especially those with visual hallucinations [25, 26]. These findings, together with our results, support an important role of cholinergic dysfunction in visuospatial impairment. Importantly, previous longitudinal studies have demonstrated that the baseline Ch4 atrophy is associated with future cognitive impairment and can predict the subsequent development of dementia in PD [8, 27]. In parallel, the PD‐PC subtype has been reported to exhibit a higher risk of progression to dementia than other PD‐MCI subtypes [28]. Given the association between Ch4 volume and visuospatial performance observed in the present study, together with previous evidence linking visuospatial dysfunction to dementia progression in PD, our findings suggest that Ch4‐related cholinergic degeneration may be a key mechanism contributing to the increased dementia risk in the PD‐PC subtype. In contrast, PD‐FS patients exhibited selective atrophy of the caudate nucleus relative to HC. Previous studies have demonstrated that dopaminergic denervation of the caudate nucleus is closely associated with executive and attentional dysfunction, supporting our interpretation that caudate degeneration is linked to the PD‐FS subtype. However, conclusions regarding the association between caudate degeneration and cognitive function in PD remain inconsistent across the literature. Our findings suggest that such discrepancies may, at least in part, be attributable to the heterogeneity of PD‐MCI subtypes. Consideration of cognitive phenotypic variability may therefore be essential for reconciling prior conflicting results concerning caudate involvement in PD‐related cognitive impairment. In addition, compared with HC, PD‐MS showed combined atrophy of the Ch4, the putamen and the thalamus, suggesting concurrent degeneration of both dopaminergic and cholinergic systems. PD‐MS patients exhibit combined impairments in attention/working memory and/or executive functions together with deficits in visuospatial functions, episodic memory, and/or language, representing a more advanced and diffuse pattern of cognitive impairment. Neuropathological studies indicate that dopaminergic and cholinergic deficits progress with disease severity [5, 29], supporting the interpretation that PD‐MS reflects a more globally degenerative phenotype. Emerging evidence indicates that dopaminergic and cholinergic systems may exert both independent and interactive effects on cognitive function in PD [30]. A recent study suggests that striatal dopamine depletion and cholinergic BF volume atrophy may interact synergistically to modulate brain metabolism and cognitive performance [31]. By delineating distinct patterns of neurodegeneration across PD‐MCI subtypes, our study provides new insights into the biological heterogeneity of cognitive impairment in PD. Previous structural MRI studies have largely treated PD‐MCI as a single clinical entity and did not specifically investigate degeneration within neurotransmitter‐related systems. In contrast, the present study examined subtype‐specific structural alterations in PD‐MCI from both dopaminergic and cholinergic perspectives. Taken together, these findings emphasize that PD‐related cognitive impairment arises not from dysfunction in a single neurotransmitter system, but rather from complex, multi‐system neurodegenerative processes. In this study, no significant differences in dopaminergic‐ or cholinergic‐related structural alterations were observed between the PD‐NC group and the different PD‐MCI subtypes. Several factors may account for this finding. First, the relatively small sample size within each PD‐MCI subtype may have limited the statistical power to detect subtle between‐group differences. Second, cognitive impairment in PD is increasingly recognized as a continuous and progressive process rather than a discrete categorical state. Consequently, some individuals classified as PD‐NC may already harbor early neurodegenerative changes that overlap with those observed in PD‐MCI, thereby reducing the sensitivity of cross‐sectional comparisons between cognitive subgroups. Longitudinal studies with larger sample sizes are warranted to further clarify the temporal evolution of these subtype‐specific structural alterations. Several limitations of this study should be acknowledged. First, volumetric measures of the basal forebrain and basal ganglia serve as indirect markers and may not fully capture underlying cholinergic or dopaminergic degeneration. Therefore, interpretations of structural alterations should be made with caution. Future studies incorporating molecular imaging techniques may help to further elucidate the consequences of basal forebrain and basal ganglia degeneration. Second, the sample size, particularly within PD‐MCI subgroups, was relatively limited, which may constrain the statistical power for subgroup comparisons. Consequently, the present findings warrant validation in larger cohorts. Third, this study was based on a cross‐sectional design, which precludes causal inferences regarding the relationship between structural alterations and cognitive impairment in PD. Longitudinal studies are therefore needed to investigate the stability and evolution of PD‐MCI subtypes and clarify the temporal progression of basal forebrain and basal ganglia degeneration and their relevance to cognitive impairment in PD. In conclusion, distinct PD‐MCI phenotypes exhibited differential patterns of dopaminergic‐ and cholinergic‐related structural alterations on MRI. PD‐MCI patients with posterior cortical impairment primarily showed atrophy of the basal forebrain, whereas those with frontostriatal deficits exhibited selective caudate nucleus atrophy. These findings provide neuroimaging‐based evidence for subtype‐specific neural substrates underlying cognitive heterogeneity in PD and highlight the distinct contributions of dopaminergic and cholinergic degeneration to different phenotypes in PD‐MCI. Author Contributions Xueqin Bai: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, data curation, supervision, resources, project administration. Haoxin Chen: methodology, investigation, visualization, validation, writing – original draft, writing – review and editing, project administration, formal analysis, software, data curation, supervision, resources. Qiuyue Li: methodology, software, data curation, supervision, formal analysis, validation, investigation, visualization, project administration, resources, writing – review and editing, writing – original draft. Shiwei Zhang: methodology, validation, investigation, writing – original draft, writing – review and editing, visualization, software, formal analysis, project administration, data curation, supervision, resources. Shuangli Chen: methodology, validation, investigation, writing – original draft, writing – review and editing, visualization, software, formal analysis, project administration, data curation, supervision, resources. Andan Qian: investigation, methodology, validation, writing – original draft, writing – review and editing, visualization, software, formal analysis, project administration, data curation, supervision, resources. Ronghui Zhou: investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation, supervision, resources. Yitong Cheng: investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, software, project administration, formal analysis, data curation, supervision, resources. Chenglong Xie: writing – original draft, investigation, methodology, validation, visualization, writing – review and editing, project administration, formal analysis, data curation, supervision, resources. Zhen Wang: investigation, writing – original draft, methodology, validation, visualization, writing – review and editing, formal analysis, project administration, data curation, supervision, resources. Yunjun Yang: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, project administration, data curation, supervision, resources. Meihao Wang: conceptualization, investigation, funding acquisition, writing – original draft, methodology, validation, visualization, writing – review and editing, software, formal analysis, data curation, supervision, resources, project administration. Funding This work was supported by General scientific research project of Zhejiang Education Department (Y202352087), the Key Research and Development Program of Zhejiang Province (2025C02111); Wenzhou Municipal Science and Technology Bureau (Y2023190), the National Natural Science Foundation of China (Grant No. 82471955), the Key Laboratory of Intelligent Medical Imaging of Wenzhou (No. 2021HZSY0057), and the Key Laboratory of Novel Nuclide Technologies on Precision Diagnosis and Treatment & clinical Transformation of Wenzhou City (No. 2023HZSY0012). Conflicts of Interest The authors declare no conflicts of interest. Supporting information Figure S1: Schematic diagram of the basal forebrain. Figure S2: Schematic diagram of the basal ganglia.