Beyond the brain: a multisystem review of pathogenesis, comorbidity, and therapeutic progress in sporadic Parkinson’s disease

Genetic factors

Research indicates that genetic factors play a pivotal role in sporadic PD. To date, over 20 high penetrance monogenic variants have been identified as potential causes of familial PD, while more than 100 genetic risk loci are associated with sporadic PD^[10-13]. During neurodevelopment, specific PAGs such as OTX2^[14-16], PITX3^[17-24], and NURR1 (NR4A2)^[25-28] are highly expressed and significantly influence the formation of the DA system^[29]. In adulthood, these genes continue to maintain neuronal homeostasis^[28-30]. Additionally, other genes exert sustained effects on DA neuron development and function throughout both developmental and adult stages. For instance, studies have categorized known PD genes into two modules based on their expression patterns during early and late brain development: the first module shows significantly higher expression in adulthood compared to the embryonic stage, while the second module exhibits the opposite pattern, with most PD genes belonging to the first module^[12]. Interestingly, patients with PD caused by mutations in the first module (highly expressed in adulthood) tend to develop the disease at a younger age than those with mutations in the second module (highly expressed during development)^[12]. This suggests that mutations affecting genes active in adulthood may be more likely to lead to early-onset disease.

During neurodevelopment, certain susceptibility genes regulate the formation and survival of mDA neurons. For instance, transcription factors such as NURR1, LMX1A/B, and EN1 mediate DA neuron differentiation during embryogenesis^[28-33]. Insufficient expression of these genes may reduce congenital DA neuron reserves, thereby increasing the risk of developing PD. NURR1 is not only crucial for early DA neuron differentiation but also maintains expression in adulthood, making it a potential therapeutic target for PD^[28,29]. Reduced NURR1 activity in PD patients correlates with substantia nigra neuron loss and inflammation, suggesting that enhancing NURR1 activity could provide protective benefits^[28,29,34]. Mice with Engrailed-1 (En1) single-allelic deletion spontaneously develop substantia nigra DA neuron degeneration and motor dysfunction, indicating EN1’s protective role in adult brains^[30]. Cohort studies also reveal associations between EN1 gene polymorphisms and susceptibility to idiopathic PD. These findings demonstrate that developmental genes continue influencing neuronal fate in adulthood, with reduced expression levels potentially contributing to PD pathogenesis. LMX1A plays a key role in mDA system development, leading to the hypothesis that its low expression in sensory nervous systems may correlate with early symptoms like olfactory dysfunction in PD patients (see Section “A Conserved genetic network links PD and olfactory dysfunction”).

In metabolic homeostasis and cellular clearance, multiple PD genes are directly involved in mitochondrial function, protein degradation, and lysosomal pathways. For instance, PINK1 and PRKN encode key proteins in the mitochondrial quality control pathway, playing crucial roles in eliminating damaged mitochondria^[35,36]. Specifically, serine protein kinase PINK1 identifies damaged mitochondria, while E3 ubiquitin ligase Parkin mediates their autophagic clearance. Mutations in PINK1 or PRKN genes can lead to autosomal recessive early-onset PD, characterized by significant mitochondrial dysfunction and elevated oxidative stress. Research shows that these genetic mutations disrupt normal mitochondrial function, causing abnormal mitochondrial accumulation in neurons. This impairs energy production while generating harmful reactive oxygen species, delivering a dual blow that ultimately triggers PD. Another example is DJ-1, which plays a role in cellular antioxidant defense^[37]. Its mutations can also cause autosomal recessive PD. These genetic defects make neurons more susceptible to metabolic stress damage. For instance, animals lacking Parkin exhibit heightened sensitivity to mitochondrial toxin rotenone. Additionally, heterozygous mutations in the GBA gene (encoding lysosomal enzyme glucocerebrosidase) are among the strongest risk factors for PD, leading to impaired lysosomal degradation and accumulation of toxic substrates^[13]. GBA mutations also disrupt glycolipid metabolism. The disease manifests through impaired degradation of α-syn in the brain, leading to its pathological aggregation. Research indicates that GBA mutations disrupt lipid metabolism, particularly by elevating levels of critical cerebroside lipids, suggesting lipid homeostasis imbalance as a key contributor to PD^[38]. The LRRK2 gene (encoding lysine-rich repeat kinase 2) is the most frequently mutated gene in autosomal dominant PD, with its protein complex regulating lysosomal function, mitochondrial dynamics, and inflammatory responses^[39,40]. LRRK2 mutations not only affect DA synapses and autophagy pathways in the brain but are also highly expressed in peripheral immune cells, establishing a critical link between metabolic and immune mechanisms in PD. Collectively, genes associated with sporadic PD influence key pathways including mitochondrial function, protein homeostasis, and neuroinflammation through multiple biological mechanisms^{[12,13,35,36,38,39]}.

Notably, certain molecular biomarkers show promise as potential indicators for early PD diagnosis or disease severity assessment. For instance, α-syn, encoded by the SNCA gene, exhibits abnormal aggregation as the core pathological feature of PD^[41,42]. Studies have demonstrated that abnormal α-syn conformational seeds can be detected in cerebrospinal fluid, with highly positive detection rates using the RT-QuIC amplification method^[43]. This suggests the technology could identify the disease years before motor symptoms emerge^[44]. Additionally, the C05-05 probe developed by Japanese scientists specifically targets α-syn fibers, detecting pathological α-syn presence through positron emission tomography (PET) imaging in both animal models and PD patients^[45]. Another example is neurofilament light chain (NfL), a sensitive axonal injury marker. In PD, particularly in advanced stages or dementia-associated cases, elevated blood NfL levels correlate closely with disease severity and progression^[46,47]. Furthermore, variations in blood concentrations of certain gene products like DJ-1 protein are being explored as potential biomarkers reflecting cellular oxidative stress and disease activity^[37]. Although no single genetic or molecular marker can currently fully predict PD, these biomarkers demonstrate promising applications in early diagnosis and disease severity evaluation.

While many of the genes discussed above were initially identified through familial forms of PD, their biological functions extend beyond monogenic inheritance and contribute to sporadic PD pathogenesis. Classical familial PD genes - including SNCA, LRRK2, PINK1, PRKN (Parkin), and DJ-1 - harbor highly penetrant mutations that directly cause Mendelian PD through mechanisms such as protein aggregation, mitochondrial dysfunction, and oxidative stress^{[35,42,48-58]}. Among these, SNCA duplication or triplication and LRRK2 gain-of-function mutations (e.g., G2019S) represent the most common dominant causes, whereas PINK1, PRKN, and DJ-1 loss-of-function variants lead to autosomal-recessive, early-onset forms of PD^{[39,52,56,58]}. Importantly, the molecular pathways governed by these genes - mitophagy, lysosomal degradation, and proteostasis - are also dysregulated in apparently sporadic cases, implying shared mechanistic foundations^[59,60].

In contrast, several developmental and regulatory genes - such as NURR1, LMX1A/B, EN1, OTX2, PITX3, and WNT5A - do not typically carry high-penetrance mutations but influence disease susceptibility through transcriptional or epigenetic modulation of mDA neuron differentiation and maintenance^{[25,30,61-63]}. Variants or reduced expression of these factors may lower the “developmental reserve” of DA neuron, predisposing individuals to later degeneration under environmental or metabolic stress. Additionally, heterozygous GBA variants occupy an intermediate position between familial and sporadic PD, functioning as strong risk factors that accelerate α-syn aggregation and cognitive decline^[64]. These mechanisms, which are amplified in familial PD, also apply to sporadic PD development, though the latter is typically triggered by the cumulative effects of multiple minor genetic variants and environmental factors. In-depth research into monogenic PD mechanisms has provided targeted intervention strategies for sporadic PD, including developing biologics to clear α-syn, activating mitochondrial quality control, or enhancing lysosomal activity, with the goal of slowing or halting disease progression.

Environmental factors

Environmental exposure plays a significant role in the development of PD, typically influencing disease risk at the population level while interacting with the susceptibility genes at the individual level. Epidemiological studies have demonstrated that exposure to pesticides, heavy metals, and air pollutants, particularly fine particulate matter (PM_2.5) and nitrogen dioxide (NO₂), is significantly associated with an increased risk of PD^[65-70].

Aging and systemic vulnerability

Ageing is the primary risk factor for PD. The prevalence of PD rises exponentially with age: approximately 1% in individuals over 60, and reaching 3% to 5% in those over 85. Population ageing is recognized as the driving force behind the rapid increase in PD cases this century^[1,95]. Aging increases the vulnerability of the central nervous system through multiple changes, vividly described as the background amplifier of PD^[96].

Environmental pressure and genetically constrained resilience

Sporadic PD arises from the interplay of sustained environmental pressure, genetically constrained resilience, and age-related erosion of compensatory capacity. As detailed in Section “Genetic Factors”, “Environmental Factors” and “Aging and Systemic Vulnerability”, genetic architecture sets the baseline robustness of molecular and cellular networks, environmental exposures impose chronic, cumulative stress, and aging progressively diminishes the capacity to buffer these insults. Disease onset thus reflects the point at which environmental load exceeds the tolerance limits defined by genetic and developmental resilience.

At the population level, environmental and lifestyle factors act as the primary external risk factors of PD risk. The exposome concept underscores that lifelong, cumulative non-genetic influences - chemical exposures, microbial shifts, metabolic stress, infections, and behavioral patterns - shape overall disease burden more than isolated events^[112]. Epidemiological evidence consistently links PD to pesticides, air pollution, and gut-related factors, reinforcing the role of environmental pressure in initiating stress across peripheral and central systems.

Yet exposure alone cannot account for the striking inter-individual heterogeneity in sporadic PD. Individuals with comparable environmental histories often exhibit divergent outcomes, from lifelong resilience to early neurodegeneration. This variability stems from differences in intrinsic resilience - shaped by genetic architecture, developmental reserve, and eroded over time by aging. Population studies integrating polygenic risk scores (PRS) with exposure data confirm that genetic susceptibility modulates environmental risk magnitude, with high PRS and high environmental burden conferring substantially greater risk than either factor alone^[113,114].

Aging serves as a critical temporal modifier, progressively narrowing resilience windows through declines in mitochondrial function, proteostasis, immune regulation, and vascular integrity. In this view, aging does not initiate PD but lowers the threshold at which cumulative environmental pressure destabilizes vulnerable networks.

Epigenetic mechanisms provide a key interface, enabling environmental exposures to induce durable changes in DNA methylation, histone modification, and non-coding RNA regulation - thereby encoding exposure history into gene expression programs without altering DNA sequence^[115,116]. Genetic background and aging further shape this epigenetic responsiveness, helping explain divergent outcomes from similar exposures.

Collectively, environmental pressure supplies the primary external load, genetic architecture defines tolerance thresholds, and aging erodes these thresholds. This dynamic framework accounts for disease heterogeneity, incomplete penetrance, and multisystem involvement in sporadic PD, while laying the groundwork for the network analyses in subsequent sections.

Immune dysregulation and autoimmunity

Immune dysregulation, particularly chronic inflammation, markedly exacerbates neurodegeneration in PD. From the adaptive immunity perspective, genome-wide association studies (GWAS) have consistently linked variants within the major histocompatibility complex (MHC) class II region - including HLA-DQB1 and HLA-DRB5 - to PD susceptibility, implicating antigen presentation and T-cell–mediated immune mechanisms in disease risk^[117].

α-Syn spreads in a prion-like manner through neuron-to-neuron transmission, with microglia playing a key facilitating role. Consistent with these genetic associations, α-syn-specific T-cell reactivity has been detected in the peripheral blood of individuals with preclinical and early PD, supporting a role for autoreactive adaptive immune responses in the disease process^[118,119].

In parallel, innate immune activation within the central nervous system plays a pivotal role in PD pathogenesis. Extracellular α-syn aggregates are efficiently sensed by microglia through pattern-recognition receptors, particularly TLR2 and TLR4, triggering robust microglial activation and inflammatory signaling^[120-123]. While microglia internalize α-syn species, this process simultaneously promotes neuroinflammatory responses that contribute to DA neuronal vulnerability. Downstream of Toll-like receptor (TLR) signaling, α-syn-induced activation of the NLRP3 inflammasome leads to the maturation and release of pro-inflammatory cytokines such as IL-1β, thereby amplifying neuroinflammation and DA neurodegeneration^[124,125]. As highlighted by Wang et al. (2015), sustained neuroinflammation and microglial activation constitute a central pathogenic mechanism in PD, providing a permissive immune environment in which α-syn–driven pathology can be amplified^[126]. Collectively, these findings support a self-reinforcing immune–α-syn axis in PD, in which adaptive immune susceptibility, innate immune activation, and microglia-mediated neuroinflammation converge to drive disease progression - a framework that highlights autoimmune-like features in sporadic PD.

To weave together the roles of genetics, environment, and aging in sporadic PD, we suggest classifying cases into three subtypes based on dominant drivers. First, environment-driven PD arises when intense exposures - like heavy pesticides or pollution - overpower even resilient genetic networks, leading to disease in people who might otherwise stay healthy^[65,69,127]; this echoes Ascherio et al.’s (2006) emphasis on environmental dominance^[128]. Second, gene network-driven PD - our review’s core focus - occurs when vulnerable conserved modules (e.g., SNCA, PINK1, PRKN) make individuals susceptible even to everyday or mild stressors. Third, mixed-driver PD involves two factors exerting roughly equal influence, such as genetics paired with environment or aging. This model recognizes the limits of our gene-focused lens for purely exposure-based cases, yet stresses genetics as the primary internal force in most sporadic PD, shaped by external elements.

A conserved genetic network links PD and olfactory dysfunction

The sensory system in PD becomes affected earlier than typical motor symptoms, with olfactory dysfunction being particularly prominent. Epidemiological and longitudinal studies show that 70%-90% of patients develop hyposmia or anosmia, often 4-10 years before motor onset, supporting an “olfactory-first” or bottom-up subtype in at least a subset of cases^{[129,130,146-148]}. We speculated a core genetic network are shared between mDA neurons and OB neurons. To validate this hypothesis, we analyzed single-cell RNA sequencing (scRNA-seq)-derived DEGs through a comparison between PD and control mDA neurons^[136], and bulk RNA-seq DEGs in OB between PD patients and controls^[140]. We identified that 712 DEGs are shared in PD mDA and OB tissues. These genes showed enrichment in Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways related to PD, Non-alcoholic fatty liver disease, Prion disease, Alzheimer disease, Oxidative phosphorylation and Huntington disease [Figure 1A]. By integrating human developmental mDA neuron scRNA-seq data^[137] (no human developmental OB transcriptome data being available), we found that 528 of the 712 genes are expressed from development through adulthood in mDA neurons. Among these, the top enriched KEGG pathway remained the PD pathway [Figure 1A]. These results suggest the existence of a core genetic network that may drive the co-occurrence of PD and olfactory dysfunction along aging. Since mouse models are commonly used to investigate PD mechanism, we also analyzed the transcriptome data from mouse models^[137-139]. We identified 4,468 and 8,570 genes shared between mDA and OB neurons in adult and developmental-stage mouse, respectively. Similar to findings in human, these genes were enriched in KEGG pathways such as Ribosome, PD, Huntington disease and Spliceosome, among others [Figure 1B]. Among them, 3,918 genes are co-expressed in mouse mDA and OB throughout development and adulthood, and are also enriched in KEGG pathways such as Ribosome, PD and Huntington disease [Figure 1B]. This recurrent enrichment across independent datasets and stages indicates that mouse mDA and OB neurons share a PD-related molecular program, suggesting that olfactory structures are embedded within a common vulnerability network underlying progressive nigrostriatal degeneration.

Figure 1. Integrated analysis of shared molecular programs between mDA neurons and the OB. (A) Venn diagram illustrates the overlap of gene sets derived from adult human DEG from mDA neurons (PD vs. control), developing human mDA neurons, and DEGs from OB samples (PD vs. control). KEGG pathway enrichment was conducted separately for genes shared only in the adult stage, only in the developmental stage, and those shared across both stages; (B) Venn diagram illustrates the overlap of gene sets derived from adult mouse mDA neurons, developing mouse mDA neurons, adult mouse OB and developing mouse OB. KEGG pathway enrichment was conducted separately for genes shared only in the adult stage, only in the developmental stage, and those shared across both stages; (C) Venn diagram illustrates the overlap of gene sets derived from human mDA vs. OB and mouse mDA vs. OB. KEGG pathway enrichment was also conducted for common gene set. The top6 enriched pathways are visualized based on signal strength, gene count, and FDR. Human adult OB transcriptomic data with sufficient quality were not available and are therefore not included in this comparison. mDA: Midbrain dopaminergic; OB: olfactory bulb; DEG: differentially expressed gene; PD: Parkinson’s disease; KEGG: Kyoto Encyclopedia of Genes and Genomes; FDR: false discovery rate.

To further test whether this mDA–OB vulnerability module is evolutionarily conserved rather than dataset-specific, we integrated human and mouse gene sets derived from shared signatures. By intersecting human developmental and adult mDA- and OB-related gene sets (with no developmental OB data available in human) with corresponding mouse developmental/adult mDA and OB datasets, we found that a conserved cross-species module of 840 genes co-expressed in mDA and OB neurons in both species [Figure 1C]. KEGG enrichment analysis of this conserved gene set again revealed significant over-representation of the PD pathway [Figure 1C], indicating that the shared mDA–OB network constitutes a PD-relevant, evolutionarily conserved vulnerability scaffold. These findings strongly support for the use of mouse models as biologically meaningful surrogates for investigating human susceptibility to PD and olfactory dysfunction.

Within the conserved mDA-OB module, several well-established PD genes provide concrete anchors supporting the biological relevance of this shared network. SNCA is central: α-syn aggregation in the olfactory epithelium and OB, together with its capacity for prion-like propagation along olfactory tracts, provides a direct molecular substrate for prodromal hyposmia and for rostrocaudal spread of pathology^[146,147]. UCHL1 is highly expressed in olfactory receptor neurons and their axons and is widely used as a pan-neuronal or sensory marker^{[129,149,150]}. Impaired UCHL1 function compromises ubiquitin–proteasome–mediated protein turnover and can facilitate α-syn accumulation, reducing proteostatic reserve in olfactory neurons exposed to environmental toxins^[151]. Dopamine D2 receptor (DRD2) is abundant in the OB DA interneurons and plays a role in adjusting synaptic input and odor discrimination. Modulation of DRD2 signaling within OB circuits can modify olfactory performance and has the potential to improve olfactory deficits in parkinsonian experimental models. This functionally links DA signaling to early sensory impairments associated with PD^[152-154]. ATP13A2, a lysosomal P-type ATPase whose loss-of-function mutations cause early-onset parkinsonism^[155,156], and whose deficiency induces lysosomal dysfunction and α-syn accumulation^[157], is expected to be equally important for protein clearance in OB neurons; subclinical ATP12A2 insufficiency could thus contribute to early olfactory α-syn vulnerability. The convergence of these canonical PD genes within the cross-species mDA-OB overlap strongly argues that this 840-gene set captures a genuine, PD-relevant vulnerability architecture, rather than a generic developmental or sensory transcriptome. Environmental exposures interact with this conserved mDA–OB network, particularly through inhalational routes that preferentially engage olfactory structures. Airborne pollutants, including PM_2.5 and NO₂, as well as certain pesticides, have been associated with oxidative stress and mitochondrial dysfunction in olfactory and DA systems. Consistent with this, epidemiological studies by Ritz et al. (2016) demonstrated that long-term exposure to traffic-related air pollution is associated with increased PD risk, while mechanistic studies have suggested that such exposures may impair oxidative phosphorylation and related mitochondrial pathways, thereby amplifying vulnerability in the OB and connected networks^[158]. These conserved vulnerability networks likely define intrinsic limits of resilience, which may be preferentially engaged - and thresholds lowered - by environmental pressures entering via the nasal interface, potentially contributing to the early and prominent olfactory involvement observed in sporadic PD.

Shared genetic architecture underlying PD and somatosensory deficits

Beyond olfaction, the somatosensory system is frequently affected in patients with PD, with pain representing one of the most prevalent non-motor symptoms. Systematic reviews estimate that approximately 40%-85% of PD patients experience pain, most commonly musculoskeletal, dystonic, radicular or central in origin^[159-161]. Neuropathological evidence shows that Lewy-related α-syn pathology is not confined to the brain: Lewy bodies and Lewy neurites have been reported in DRG, spinal dorsal horn, and other peripheral sensory structures in PD and Lewy body spectrum disease. Population-based mapping studies confirm the presence of peripheral and spinal Lewy pathology in defined disease subtypes^[162-164].

Inspired by the conserved molecular architecture common to mDA and OB neurons, we subsequently inquired whether similar genetic vulnerability modules extend to peripheral sensory neurons, particularly those in DRG. To explore this, we intersected PD-related genes from mDA signatures with transcriptomic profiles of DRG neurons from human and mouse datasets, encompassing developmental and adult stages. Utilizing gene sets derived from developing and adult mDA neurons, including PD-relevant DA modules^[136,137], we compared their expression with curated DRG datasets enriched for sensory neurons^[141,165]. Cross-tissue intersections analysis was then performed to identify conserved and disease-relevant molecular modules. We identified 3,651 shared genes between PD mDA DEGs and adult DRG neurons. These genes showed enrichment in KEGG pathways including Ribosome, PD, Oxidative phosphorylation, Prion disease, Huntington disease and amyotrophic lateral sclerosis (ALS) [Figure 2A]. During development, 8,595 genes are co-expressed in both human mDA and DRG neurons. Among these, the PD pathway was significantly enriched, ranking among the top five KEGG pathways. By integrating human developmental and adult data, we identified a robust shared gene set of 3,089 genes co-expressed in both tissues and significantly associated with PD-related alterations, with the PD pathway ranking among the top two enriched KEGG pathways [Figure 2A]. A similar pattern was observed in mouse model^[137,141], where 3,936 genes were shared between mDA and DRG from development throughout adulthood, and the PD pathway ranked among the top three significantly enriched KEGG pathways [Figure 2B].

Figure 2. Overlapping gene networks between mDA neurons and dorsal root ganglion sensory neurons (mDA and DRG). (A) Venn diagram illustrates the overlap of gene sets derived from adult human DEG from mDA neurons (PD vs. control), developing human mDA neurons, adult human DRG and developing human DRG. KEGG pathway enrichment was conducted separately for genes shared only in the adult stage, only in the developmental stage, and those shared across both stages; (B) Venn diagram illustrates the overlap of gene sets derived from adult mouse mDA neurons, developing mouse mDA neurons, adult mouse DRG and developing mouse DRG. KEGG pathway enrichment was conducted separately for genes shared only in the adult stage, only in the developmental stage, and those shared across both stages; (C) Venn diagram illustrates the overlap of gene sets derived from human mDA vs. DRG and mouse mDA vs. DRG. KEGG pathway enrichment was also conducted for common gene set. The top6 enriched pathways are visualized based on signal strength, gene count, and FDR. mDA: Midbrain dopaminergic; DRG: dorsal root ganglia; DEG: differentially expressed gene; PD: Parkinson’s disease; KEGG: Kyoto Encyclopedia of Genes and Genomes; FDR: false discovery rate.

To evaluate whether the mDA–DRG vulnerability module represents a conserved biological program rather than dataset-specific artifacts, we performed a cross-species integration of the overlapping signatures. Human developmental and adult mDA- and DRG-associated gene sets were systematically compared with matched mouse developmental and adult mDA/DRG transcriptomes. This analysis revealed a reproducible conserved module of 7,475 genes consistently co-expressed in both mDA and DRG neurons across species [Figure 2C]. KEGG pathway analysis of this conserved subset again showed significant enrichment for the PD pathway [Figure 2C], along with Ribosome, ALS, Huntington disease, Spliceosome and Prion disease, confirming that the mDA–DRG overlap captures a PD-relevant, evolutionarily conserved vulnerability network. Together, these findings support the use of mouse midbrain and DRG models as mechanistically relevant systems for investigating shared molecular mechanisms underlying nigrostriatal degeneration and somatosensory deficits in PD.

Within the conserved mDA-DRG module, several well-established PD genes offer concrete anchors that support the biological relevance of this shared network. SCN9A, FAAH and GCH1 define a peripheral “sensory gain” module within DRG neurons that interfaces with central DA vulnerability in PD. Nav1.7 encoded by SCN9A is a key determinant of nociceptor excitability, and PD cohorts show SCN9A variants enriched in patients with refractory or severe pain, supporting a genetic lowering of pain threshold in the parkinsonian state^[166,167]. Loss-of-function or hypomorphic variation in FAAH, a major endocannabinoid-degrading enzyme, can prolong endogenous analgesic signaling and conversely, risk alleles have been associated with heightened pain sensitivity, including in PD, indicating that endocannabinoid tone shapes individual susceptibility to PD-related pain^[168,169]. GCH1, via control of tetrahydrobiopterin (BH4) and thus dopamine and nociceptive monoamine synthesis, acts as both a PD risk/modifier gene and a pain gene, with variants influencing PD risk, nigrostriatal integrity and inflammatory–nociceptive signaling, positioning BH4 metabolism at the intersection of mDA degeneration and sensory hyperexcitability^[170-172]. In parallel, pathogenic mutations in core monogenic PD genes, such as SNCA, PINK1, PARK7, ATP13A2, VPS35, DNAJC6, FBXO7, SYNJ1, RAB39B, CHCHD2, and PLA2G6, converge on mitochondrial quality control, lysosomal-autophagic flux, and synaptic vesicle/endocytic pathways in mDA neurons^[173-176]. These mutations converge on pathways governing mitochondrial quality control, lysosomal–autophagic function, and synaptic vesicle/endocytic trafficking, and are well-established contributors to nigrostriatal vulnerability in PD. When mapped together with peripheral sensory channel and neuromodulator genes such as SCN9A, FAAH, and GCH1, they delineate a partially shared gene network between PD-linked molecular pathology and DRG-based nociceptive regulation. This overlapping mDA-DRG 7475-gene module suggests a genetic interface between PD biology and pain/somatosensory pathways, rather than demonstrates a simple causal mechanism. For example, epidemiological and experimental studies have shown that exposure to pesticides and other environmental toxicants is associated with increased PD risk, with oxidative stress and mitochondrial dysfunction, overlapping with mitochondrial quality-control pathways implicated in PD, such as those involving PINK1^[65]. In parallel, airborne pollutants have been shown to elicit inflammatory responses in exposed peripheral tissues, including sensory systems, by directly activating peripheral sensory neurons and associated inflammatory signaling, thereby potentially modulating DRG-mediated vulnerability relevant to PD^[177]. These shared molecular architectures likely constrain neuronal resilience under sustained stress, providing a plausible mechanistic substrate through which environmental load contributes to somatosensory dysfunction - such as heightened pain sensitivity - and overall vulnerability in sporadic PD.

A conserved genetic network bridging central and peripheral neuronal vulnerability

The ENS embedded in the gastrointestinal tract is increasingly recognized as an early site of PD pathology. We hypothesize that the ENS shares a developmental and molecular homology with central DA structures, rendering it susceptible to parallel pathological processes under parkinsonian conditions. To test this hypothesis, we integrated and compared recent high-resolution transcriptomic datasets from the adult or developmental human enteric and mDA neurons^{[136,137,143,145]}, performing overlap analysis to identify conserved gene modules that may underlie their shared vulnerability in PD. We identified 736 DEGs that are shared in PD mDA and enteric neurons. These genes showed enrichment in KEGG pathways related to Ribosome, PD, Oxidative phosphorylation, Non-alcoholic fatty liver disease, Thermogenesis and Cardiac muscle contraction [Figure 3A]. During development, 1,594 genes are co-expressed in both human mDA and enteric neurons. Among these, the PD pathway was significantly enriched, ranking among the top two KEGG pathways. By integrating human developmental and adult data, we identified 153 genes co-expressed in both tissues and significantly associated with PD-related alterations, and the PD pathway was significantly enriched, ranking among the top two KEGG pathways [Figure 3A]. A similar pattern was observed in mouse model^[145,178], where 8,873 and 8,817 genes were shared between mDA and enteric neurons in adult and developmental stages, respectively, and the PD pathway ranked among the top four significantly enriched KEGG pathways [Figure 3B]. Among them, 7,778 genes are co-expressed from development throughout adulthood and enriched in PD pathway [Figure 3B].

Figure 3. Shared transcriptional modules between mDA neurons and the ENS. (A) Venn diagram illustrates the overlap of gene sets derived from adult human DEG from mDA neurons (PD vs. control), developing human mDA neurons, adult human ENS and developing human ENS. KEGG pathway enrichment was conducted separately for genes shared only in the adult stage, only in the developmental stage, and those shared across both stages; (B) Venn diagram illustrates the overlap of gene sets derived from adult mouse mDA neurons, developing mouse mDA neurons, adult mouse ENS and developing mouse ENS. KEGG pathway enrichment was conducted separately for genes shared only in the adult stage, only in the developmental stage, and those shared across both stages; (C) Venn diagram illustrates the overlap of gene sets derived from human mDA vs. ENS and mouse mDA vs. ENS. KEGG pathway enrichment was also conducted for common gene set. The top6 enriched pathways are visualized based on signal strength, gene count, and FDR. mDA: Midbrain dopaminergic; ENS: enteric nervous system; DEG: differentially expressed gene; PD: Parkinson’s disease; KEGG: Kyoto Encyclopedia of Genes and Genomes; FDR: false discovery rate.

To determine whether the mDA–ENS vulnerability module represents an evolutionarily conserved architecture rather than a dataset-specific artifact, we next integrated shared gene signatures derived from both human and mouse. Specifically, we intersected gene sets co-expressed in human mDA and enteric neurons at developmental and/or adult with corresponding mouse developmental and/or adult mDA and enteric neurons gene sets. This cross-species analysis identified a robust conserved module of 2,166 genes jointly represented in mDA and enteric neurons across species [Figure 3C]. KEGG pathway analysis of this conserved set again revealed significant enrichment of the PD pathway [Figure 3C], indicating that the mDA–ENS network reflects not merely a generic developmental or peripheral neuronal program, but a PD-relevant, evolutionarily conserved vulnerability scaffold.

Within the conserved mDA-ENS module, several well-established PD genes offer concrete anchors that support the biological relevance of this shared network. SNCA sits at the core: α-syn pathology is repeatedly detected in the ENS and other peripheral autonomic sites in PD-findings that helped inspire “gut-first” hypotheses of disease spread^{[130,179,180]}. PINK1-PRKN encode the mitophagy axis that safeguards neuronal and enteric mitochondrial quality. In the intestine, infection can trigger mitochondrial-antigen presentation and PD-like phenotypes in Pink1^-/- mice^[181], while human PD patients harbor T-cell responses to PINK1-derived mitochondrial peptides, linking this pathway directly to gut immune mechanisms relevant to PD^[182]. LRRK2 mechanistically links PD genetics to mucosal immunity: it modulates myeloid responses, and its overexpression can exacerbate experimental colitis, supporting a role for LRRK2-driven immune signaling at the gut interface, with potential implications for gut-brain communication in PD^[183]. PARK7 (DJ-1) buffers oxidative and inflammatory stress; loss of DJ-1 heightens gut mucosal inflammation and worsens colitis in vivo, consistent with a model in which impaired cytoprotection in the intestinal milieu can feed forward onto PD-relevant ENS vulnerability^[184]. This overlapping mDA-ENS 2166-gene module suggests a genetic interface brain-gut axis in PD biology, rather than demonstrates a simple hypothesis.

Convergent modules along the sensory-gut-midbrain axis

To refine the gene sets most relevant to human PD pathology, we performed intersection analysis between mDA and each of the three peripheral tissues (OB, DRG, ENS). This approach identified two distinct groups: 265 genes co-expressed exclusively in humans and 244 genes shared between humans and mice [Figure 4A]. Each gene was manually classified into one of three categories - established PD relevance, potential PD relevance, or no current evidence of PD association. Based on these results, a comprehensive PPI network was constructed [Figure 4B and C], identifying several central “hub” regulators and potential regulatory clusters.

Figure 4. Cross-species comorbidity gene overlap and functional enrichment analysis. (A) Venn diagram showing the overlap between comorbidity-associated genes identified in humans only (244 genes) and those conserved across both human and mouse datasets (265 genes); (B) PPI network of comorbidity genes identified only in humans; (C) PPI network of comorbidity genes shared between human and mouse. Divided into three clusters: Related to PD, Potentially related to PD and Genes without established association with PD. Note: This classification is based on the literature and serves as a hypothesis-generating framework rather than a definitive annotation. “Genes without established association with PD” cluster are further grouped into seven functional submodules based on biological relevance: neural development and synapses (green), autophagy and protein clearance transport (yellow), mitochondrial function and oxidative stress (brown), inflammation and immune metabolism (pink), translation and ribosome/protein synthesis (blue), basic cellular functions and cytoskeleton (purple), and RNA processing/splicing/nuclear export and transcriptional regulation(orange); (D) KEGG pathway enrichment analysis of the conserved comorbidity genes expressed in both species. Related to PD: established PD relevance; Potentially related to PD: potential PD relevance; Genes without established association with PD. PPI: Protein–protein interaction; PD: Parkinson’s disease; KEGG: Kyoto Encyclopedia of Genes and Genomes; DEG: differentially expressed gene; mDA: midbrain dopaminergic; DRG: dorsal root ganglia; ENS: enteric nervous system; OB: olfactory bulb; FDR: false discovery rate.

Among these intersection sets, several genes are well-established PD-associated molecules, including SNCA, NEAT1, NEFL, UCHL1, and ATP13A2. SNCA, encoding α-syn, remains the most validated causative gene in familial and sporadic PD. Point mutations and multiplications of SNCA drive α-syn aggregation and inter-neuronal propagation, forming the molecular basis for the “prion-like” spreading mode^[49,185]. NEAT1, a long noncoding RNA, regulates SNCA expression and neuroinflammatory responses. Elevated NEAT1 in PD patient brains and in vitro models enhances α-syn accumulation via PINK1- and NLRP3-dependent pathways, amplifying oxidative stress and inflammation^{[163,186,187]}. NfL, encoded by NEFL, is a key structural component of axons; knockout and mutant mouse studies demonstrate that loss or disruption of NEFL induces axonal degeneration and altered neuronal integrity^[188,189], while clinical cohorts consistently show elevated CSF and blood NfL in PD and other parkinsonian syndromes, with higher levels associated with more severe disease and faster progression^[190-193], establishing NEFL/NfL as a robust marker of axonal damage and prognosis. Ubiquitin C-terminal hydrolase L1 (UCHL1), is a neuron-enriched deubiquitinase essential for maintaining the free ubiquitin pool and proteasomal function; rare PD-associated mutations (such as I93M) and functional studies implicate UCHL1 dysfunction in impaired protein clearance, α-syn aggregation, and axonal vulnerability, positioning it as a proteostasis-sensitive node within PD-relevant circuits^[194-196]. ATP13A2 (PARK9) was discovered as the gene mutated in Kufor–Rakeb syndrome, an autosomal recessive juvenile-onset parkinsonism^[155], and Atp13a2-deficient models develop lysosomal storage pathology, impaired autophagy, and sensorimotor deficits^{[157,197,198]}. Cell and animal studies further demonstrate that ATP13A2 loss-of-function causes generalized lysosomal dysfunction and promotes α-syn accumulation and neurodegeneration^[199,200], firmly linking this lysosomal P-type ATPase to PD pathogenesis.

Genes without established association with PD were further grouped into seven functional categories - neural development and synapses, autophagy and protein clearance transport, mitochondrial function and oxidative stress, inflammation and immune metabolism, translation and ribosome/protein synthesis, basic cellular functions and cytoskeleton, and RNA processing/splicing/nuclear export and transcriptional regulation - to serve as a reference resource for hypothesis-driven exploration [Figure 4B and C]. Within the “Genes without established association with PD” subgroup, several genes nevertheless suggest plausible PD-related mechanisms. In the human-only set, SYNGR3 encodes a synaptic vesicle membrane protein that interacts with the dopamine transporter (DAT) and shapes presynaptic dopamine handling; overexpression of Synaptogyrin-3 enhances striatal dopamine uptake and partially rescues DA dysfunction in LRRK2^R1441G mutant mice, pointing to a potential role as a synaptic modifier of PD vulnerability^[201]. RTN4 (Nogo-A) is an ER membrane protein that shapes tubular ER and axons. RTN4 levels in CSF are elevated in neurodegenerative diseases including PD, and Nogo-A-targeting strategies improve DA neuron survival in PD models, suggesting a link between ER/axon integrity and PD pathology^[202]. Among human-mouse common genes, TMEM163 encodes a zinc-transporting membrane protein. GWAS and follow-up studies implicate TMEM163 as a PD risk locus and associate rare variants with early-onset PD, supporting a role for disturbed synaptic Zn²⁺ homeostasis in PD^[203]. Finally, SNCG encodes γ-synuclein, a synuclein family member capable of aggregation. Genetic and pathological data indicate that SNCG variants and γ-synuclein-positive inclusions contribute to synucleinopathy risk, including PD and dementia with Lewy bodies^[204]. Beyond these highlighted genes, our “Genes without established association with PD” module also contains some plausible candidates that converge on pathways already implicated in PD pathogenesis. IFNGR1, GRN, GDF11 and MAP4K4 participate in immune and stress-response signaling, whereas TOMM40, FTL and NPC1 are involved in mitochondrial, lysosomal and iron-handling pathways that are critical for neuronal resilience. GAP43, THY1, SV2A, GRIK2, RGS12, CD63, SORT1 and GPRC5B are linked to synaptic plasticity, vesicle and exosome trafficking and receptor signaling, suggesting that subtle dysregulation of these nodes may further shape network-level vulnerability in sporadic PD.

Rather than proposing a definitive mechanistic model, this integrative cross-species analysis provides a structured reference framework for future studies. By mapping conserved and tissue-specific molecular overlaps across sensory, gut, midbrain, and peripheral systems, our findings highlight candidate genes and pathways that warrant targeted experimental validation. The inclusion of both well-established PD-related genes (e.g., SNCA, NEAT1, NEFL, UCHL1, and ATP13A2) and previously uncharacterized “Genes without established association with PD” genes offers a prioritized landscape of potential molecular nodes within this multisystem network [Figure 4]. Building on this framework, the subsequent sections of this chapter dissect major clinical and pathological domains, sensory dysfunction, gut–brain axis alterations, and systemic metabolic and immune changes, to assess how existing evidence aligns with and reinforces the cross-system modules identified above. Environmental factors influence this conserved mDA–ENS network, particularly through chronic metabolic, microbial, and inflammatory pressures that selectively challenge gut vulnerability. For instance, pesticides and heavy metals ingested via water or food impair autophagy and mitochondrial pathways in genes such as LRRK2 and PINK1-PRKN, as evidenced by Tanner et al. (2011) and Ascherio et al. (2006)^[128]. Similarly, gut dysbiosis and infection-driven inflammation induce mitochondrial stress and disrupt mitophagy and proteostasis - processes that overlap with the PD-enriched mitochondrial and lysosomal modules identified here. These shared vulnerability networks position the gut as a key interface where environmental load intersects with genetically constrained resilience, potentially exacerbating gut-brain axis vulnerability and contributing to early non-motor manifestations in sporadic PD.

Systemic amplifiers: metabolic and immune dysfunction

In the pathogenesis of PD, systemic metabolic status and immune system function act as “amplifiers” that can accelerate or exacerbate central nervous system pathology. First, metabolic disorders and PD share a bidirectional relationship. Epidemiological studies indicate that metabolic syndromes such as type 2 diabetes increase both the risk and progression rate of PD. For instance, a 2021 meta-analysis showed that type 2 diabetes is associated with a 27% higher risk of developing PD^[65,205,206]. Additionally, medications that improve insulin resistance, such as glucagon-like peptide-1 receptor agonists (GLP-1RA), not only help lower blood sugar but may also enhance DA neuron function, thereby reducing PD risk or improving symptoms; observational cohorts link GLP-1RA or DPP-4 inhibitor use to lower incident PD in diabetes^[207-209]. A typical example is GLP-1RA: randomized trials in PD show signal for disease-modifying benefit with lixisenatide^[210], while exenatide had earlier phase-2 benefit but a later phase-3 did not meet its endpoints, underscoring ongoing uncertainty^[211,212]. Meanwhile, preclinical/early translational work supports GLP-1RA neuroprotection and anti-inflammatory effects^[213]. Its mechanisms involve multiple aspects, including enhancing systemic and brain-level insulin signaling, mitochondrial support, and suppression of inflammatory pathways; AMPK activation with downstream mTOR inhibition promotes autophagy and mitophagy, a pathway engaged by agents like metformin in DA models^[214]. The AMPK-mTOR pathway, serving as a core mechanism for cellular energy sensing, plays a pivotal role in PD pathology. Under conditions of nutritional excess or abnormal insulin signaling, excessive mTOR activation inhibits autophagy, fostering toxic protein accumulation; conversely, AMPK activation (e.g., by metformin) suppresses mTOR, promotes autophagy/mitochondrial renewal, and protects mDA neuron in vivo^[214]. Parkinson’s patients also show characteristic brain glucose-metabolism patterns on FDG-PET (the PD-related metabolic network), which correlate with disease stage and can aid differential diagnosis^[215]. Some researchers therefore frame a shared “metabolic-neurodegenerative” axis between diabetes and PD.

Secondly, immune dysregulation, particularly chronic inflammatory states, is considered to significantly exacerbate the neurodegenerative process in PD. Evidence of immune system activation is commonly observed in PD patients, including elevated levels of pro-inflammatory cytokines (such as TNF-α and IL-1β) in peripheral blood compared to healthy individuals, as well as reactive proliferation and phagocytic phenotype of microglia in the central nervous system. Genetic studies further validate immune involvement in PD: Large-scale GWAS have identified significant associations between genetic variants encoding Class II major histocompatibility complex (MHC II) molecules (e.g., HLA-DQB1, HLA-DRB5) and PD susceptibility^[117]. This suggests that antigen-presenting and T-cell-mediated autoimmune processes may be implicated in PD. Recent research has revealed the presence of autoreactive T cells targeting neuroantigens such as α-syn in PD patients, with increased numbers of these cells in peripheral blood that may cross the blood-brain barrier to infiltrate the central nervous system and participate in aggressive attacks on DA neuron^[216]. These adaptive immune abnormalities add a hidden dimension to PD as an autoimmune-like condition, leading some scholars to propose that autoimmune mechanisms may contribute to disease progression, rather than classifying PD as a classical chronic autoimmune disorder.

More specifically, the neuroinflammatory response observed in PD brains amplifies neuronal damage through a feedback loop. Aggregated α-syn itself serves as a potent stimulus for immune responses - studies have shown that α-syn oligomers can activate TLRs, particularly TLR2 and TLR4, on microglia surfaces^[122,123]. Activation of these receptors triggers the downstream NF-κB signaling pathway, inducing extensive gene transcription that leads to the release of pro-inflammatory cytokines such as IL-1β and TNF-α^[217]. These cytokines can directly impact DA neurons, contributing to mitochondrial dysfunction and apoptotic signaling. Simultaneously, they induce neighboring astrocytes to adopt an A1 harmful phenotype, resulting in loss of neurotrophic support and increased production of toxic substances. Additionally, NF-κB upregulates and activates the NLRP3 inflammasome, leading to the maturation and amplification of inflammatory cascades like IL-1β^[218,219]. Importantly, this feedback loop primarily functions to amplify ongoing neuronal stress and proteinopathy, rather than acting as an independent initiating cascade. In summary, during PD neurodegeneration, immune activation and inflammatory factors form a positive feedback loop: initial neuronal damage (potentially caused by toxic proteins or environmental toxins) triggers glial immune responses^[220]. The released inflammatory mediators further damage neurons, expose more autoantigens, and recruit additional immune cells, perpetuating this vicious cycle that progressively worsens the disease^[124].

Recent studies indicate that PD may result from the interaction of multiple factors including genetics, environment, metabolism, and immunity. Abnormalities in the immune system, such as associations between specific immune genes and PD, suggest immune factors play a role in disease progression. Therefore, the amplifying effects of metabolic and immune factors on PD further emphasize the necessity of viewing PD as a multifactorial outcome. Some researchers propose adding a third “hit” - metabolic/immune abnormalities - as a modulatory layer to Braaks “double hit” hypothesis, forming the “triple hit” hypothesis to comprehensively explain PDs pathogenesis. In summary, while the pathological pathology of α-syn protein and the vulnerability of DA neuron are central to PD, their progression rate and severity are significantly influenced by systemic metabolic status and immune environment. These “amplifiers” not only exacerbate the pathological burden of PD but also represent critical factors that require focused consideration in future comprehensive prevention and treatment strategies.

Taken together, these findings position immune dysregulation and α-syn–associated inflammatory responses not as independent initiators of PD, but as potent systemic amplifiers that exacerbate neuronal vulnerability and accelerate disease progression in the context of genetic susceptibility and environmental pressure.

A systematic model of multisystem vulnerability in PD

Based on the evidence reviewed above regarding PD-associated olfactory dysfunction, somatosensory deficits, central–peripheral neuronal vulnerability, and the sensory–gut–midbrain axis, PD can be understood as a cross-system disorder shaped by sustained environmental pressures acting upon genetically constrained vulnerability networks across sensory, intestinal, and midbrain systems^[221,222]. Rather than reflecting a single-site lesion or a unidirectional propagation process, the cross-system networks identified here delineate shared architectures of susceptibility that define where pathological stress is most likely to accumulate under sustained challenge. Peripheral sensory circuits, such as olfactory and somatosensory pathways, and the ENS may represent the earliest sites of pathological stress, where environmental or endogenous pressures impose cumulative stress that promotes protein misfolding, mitochondrial impairment, and chronic inflammation^[223]. These local disturbances gradually converge on the mDA system through neural, immune, and metabolic routes, amplifying neurodegenerative cascades via systemic feedback loops^[224]. Importantly, such pressures do not act as isolated triggers but interact with pre-existing network constraints that shape individual tolerance and adaptive capacity. Within this framework, mDA degeneration emerges not as the inevitable endpoint of peripheral pathology, but as part of a coupled, multisystem process in which vulnerable networks across peripheral and central domains undergo parallel destabilization. Neural, immune, and metabolic connectivity enables stress signals to integrate across organs, allowing local failures to amplify one another through feedback loops rather than simple anatomical spread.

Genetic architecture defines the structure and robustness of these vulnerability networks, while aging progressively erodes compensatory mechanisms and lowers activation thresholds across interconnected systems^[225]. Age-related declines in mitochondrial bioenergetics, proteostasis, immune resolution, and vascular integrity narrow the margin between physiological stress and pathological failure, providing a mechanistic basis for the delayed onset, incomplete penetrance, and heterogeneous clinical trajectories characteristic of sporadic PD.

Critically, the interaction between genetic architecture and environmental exposure is likely multiplicative rather than additive^[226]. Genetic networks establish baseline resilience and compensatory capacity across tissues, whereas cumulative environmental exposures - toxicants, metabolic stress, microbial perturbations, and inflammatory burden - determine the magnitude and duration of imposed stress. Gene–environment interactions (G×E) thus shape individual disease trajectories by modulating vulnerability thresholds rather than acting independently. This model also accounts for incomplete penetrance and heterogeneity in sporadic PD, where many genetically at-risk individuals remain disease-free in the absence of sufficient environmental load. Consequently, conserved genetic programs primarily delineate susceptibility architecture, while human-specific exposures dictate whether, when, and where this vulnerability manifests clinically.

In this environment-influenced, genetically constrained framework, multisystem involvement reflects coordinated loss of resilience across sensory, intestinal, metabolic, and immune domains, with systemic amplifiers intensifying the impact of cumulative environmental load. Rather than proposing a definitive mechanistic pathway, this integrative model serves as a conceptual reference for understanding prodromal features, identifying early biomarkers, and guiding therapeutic strategies that extend beyond DA replacement toward restoring molecular homeostasis across the sensory–gut–midbrain axis.

Building upon these insights, the next section focuses on current and emerging therapeutic strategies, tracing the evolution from symptomatic relief to disease modification and exploring how systemic mechanisms can inform multi-target, cross-organ interventions for PD.

Existing drugs and standard therapies

Emerging disease-modifying therapies

Given the limitations of symptomatic treatments, current research in PD is shifting toward interventions that directly target the molecular and cellular drivers of neurodegeneration. These emerging approaches reflect PD’s multifactorial nature - encompassing protein aggregation, genetic mutations, mitochondrial dysfunction, and loss of neuronal integrity - and aim to alter disease trajectory rather than simply alleviate symptoms. The following sections summarize major categories of disease-modifying strategies under active investigation.

Future trends of PD therapy

Existing modalities and progress

A major challenge in PD is diagnostic delay: clinical diagnosis still relies largely on the emergence of cardinal motor symptoms, which occur only after substantial degeneration of the nigrostriatal system, with estimates suggesting about 30%-50% loss of substantia nigra DA neurons and ~50%-80% depletion of striatal dopamine terminals at motor onset^{[109,130,288-290]}. Therefore, developing early diagnostic techniques to intervene when neural damage is still minimal is crucial for improving prognosis. Recent advancements across multiple fields have brought us closer to the goal of advancing the time of diagnosis by 5 to 10 years:

Imaging biomarkers: Functional imaging can detect changes in the dopamine system before clinical symptoms emerge. Dopamine transporter imaging (DAT-SPECT) utilizes radioactive tracers such as [¹²³I]-ioflupane binding to striatal DAT enables quantification of DA neuron terminal density^[291-293]. Early-stage PD can be detected through reduced uptake in one side of the striatum via PET, which aids in differentiating atypical tremor symptoms. PET imaging using tracers like [¹⁸F]-DOPA reflects decreased dopamine synthesis function, playing a crucial role in clinical diagnosis^[294]. As an emerging MRI technique, neuro-melanin-sensitive magnetic resonance imaging (NM-MRI) holds significant diagnostic value for PD. In normal midbrain substantia nigra, abundant neuro-melanin produces high signal intensity on T1-weighted sequences. However, in PD patients, this high signal band markedly diminishes or disappears. Studies demonstrate that NM-MRI achieves high accuracy in distinguishing PD patients from healthy elderly individuals using 7T ultra-high field MRI technology, making it a promising non-invasive early diagnostic tool^[295,296]. Additionally, MRIs iron load imaging (SWI or Quantitative Susceptibility Mapping) detects iron deposition in substantia nigra, which increases in early PD patients and shows negative correlation with reduced NM signals^[297-300]. MRI techniques such as subcortical nucleus volume measurement and brain network connectivity analysis also provide robust support for related research. Overall, the imaging field is progressively advancing. The diagnostic approach is evolving from exclusionary methods (such as ruling out stroke or tumors) to quantitative detection of early PD pathology. With advancements in high-resolution MRI technology and specialized sequences, it is anticipated that a single MRI scan could screen high-risk PD candidates [e.g., individuals with anosmia combined with resting-state brain dysfunction (RBD) symptoms]^[297,300]. Integrating imaging techniques with AI analysis further enhances diagnostic sensitivity and specificity^[301].

Body fluid biomarkers: An ideal PD biomarker should originate from easily accessible samples while demonstrating high sensitivity and specificity^[302]. The most promising research direction currently is detecting abnormal conformations of α-syn. Leveraging the self-propagating characteristic of pathological α-syn, scientists have developed SAA techniques encompassing RT-QuIC and PMCA methods^[283,303]. The procedure involves adding patient samples (cerebrospinal fluid or peripheral tissue homogenates) to normal α-syn protein solution. If trace amounts of abnormal α-syn seeds are present, they induce aggregation of normal α-syn, which can be detected by fluorescent probes (e.g., ThT). Multicenter studies show that RT-QuIC demonstrates ~85%-95% sensitivity and ~90%-96% specificity in cerebrospinal fluid PD detection, particularly exhibiting exceptionally high positivity rates for patients with RBD^[43,304,305]. To reduce invasiveness, skin and nasal/olfactory mucosa samples have been explored: skin RT-QuIC and skin phospho-α-syn IHC show high specificity and high positivity in PD/synucleinopathies in prospective multicenter studies^[304,305]. Another highly promising biomarker is NfL, a protein found in the axonal skeleton that leaks into cerebrospinal fluid and bloodstream when brain neurons are damaged. While early changes in NfL levels during PD may not be as pronounced as in other neurodegenerative diseases like ALS, recent studies show that elevated plasma NfL levels are closely linked to rapid disease progression and cognitive decline risks in PD patients. This makes NfL a potential key biomarker for monitoring PD’s progression^[46,306]. Other biomarkers such as DJ-1 protein, tau protein, and inflammatory factors have been reported, but they lack specificity and consistency. Currently, multiple institutions are collaborating to improve diagnostic accuracy through biomarker combinations. For instance, the PPMI study suggests combining α-syn RT-QuIC results, saliva-based miRNAs regulating DJ-1 gene expression, and urine NfL testing for better predictive power. In summary, the field of biomarkers is advancing rapidly and holds promise for clinical application in the near future.

Peripheral tissue biopsies: As previously mentioned, skin biopsies serve as a window to obtain central pathological information. Similarly, colon mucosal biopsies and salivary gland biopsies are also under investigation. Results indicate that approximately 70% to 80% of PD patients show immunohistochemical detection of phosphorylated α-syn inclusions in intestinal mucosal or lower lip salivary gland biopsy specimens, while controls rarely exhibit positive findings. While these biopsies combined with highly specific antibodies demonstrate diagnostic value, their sensitivity remains limited and sampling errors may occur^{[306,307,308]}. A more minimally invasive approach involves nasal endoscopic scraping of olfactory mucosa cells, which has been applied in RT-QuIC testing. Peripheral tissue samples also hold promise for transcriptomic/metabolomic analysis to identify diagnostic biomarkers through pattern recognition. However, these methods are currently in the research phase and remain distant from widespread clinical application.

Digital phenotyping: Leveraging wearable devices, smartphone sensors, and AI algorithms, this technology captures subtle motor and behavioral changes in PD patients. For instance, wrist-mounted accelerometers track gait patterns and tremor frequency, while smartphone touchscreen tapping rhythms reflect finger dexterity. Typing error rates indicate fine motor dysfunction. Voice analysis detects weak vocalizations and slowed speech rates to identify speech disorders early. Sleep monitors can detect abnormal nocturnal movements and heart rate variations that may help screen for REM sleep behavior disorder^[309]. The advantages of digital phenotyping - continuity, objectivity, and non-invasiveness - have been validated in multiple PD studies as correlating with traditional clinical assessments. Particularly for individuals with family history or mild symptoms, digital monitoring establishes baseline data and tracks trend changes^[310]. Some companies are developing mobile Apps for self-monitoring, enabling patients to perform home-based finger tapping and voice tests, then upload data to cloud platforms for analysis. While these tools cannot yet serve as standalone diagnoses, they show great promise as clinical assessment aids and high-risk screening tools. A study published in Science Advances demonstrated that integrating keystroke dynamics from intelligent keyboard systems (IKS) with gait acceleration patterns reveals behavioral biomarkers^[311]. By analyzing degree and speech spectrum data, it can effectively distinguish patients with mild PD from healthy individuals with over 80% accuracy. In the future, digital phenotypes are expected to integrate with biomarkers to enable remote, early, and multidimensional detection of PD.

Future directions

Although multiple methods are now available for early diagnosis of PD, new approaches and tools are required to achieve more advanced prediction and personalized risk assessment. The following are some future directions:

Evaluation of Developmental Risk Factors: Although multiple methods are now available for early diagnosis of PD, new approaches and tools are required to achieve more advanced prediction and personalized risk assessment. The following directions emphasize risk identification and early-life susceptibility profiling rather than therapeutic intervention. Developmental factors may influence lifelong vulnerability to PD. For instance, variations in the number of DA neuron in the substantia nigra at birth mean individuals with lower neural “reserves” may be more prone to functional deficits in middle age^[312]. Another example: Could developmental injuries like brain inflammation or trauma during childhood potentially create hidden risks? Current cohort studies are tracking premature infants, low birth weight babies, and those exposed to maternal infections during pregnancy to observe whether their PD incidence increases in old age. Research also suggests that hormone levels around puberty and high-intensity exercise habits may shape adult brains resistance to PD^[313]. Future developments might include scoring systems based on developmental indicators - such as motor coordination in childhood, cognitive development, and family history - to predict lifelong PD risk. While long-term data validation remains necessary, this highlights the concept that “PD origins may trace back to childhood”. These findings underscore the importance of incorporating developmental indicators into long-term risk models and early surveillance frameworks.

Multi-omics integrated Risk Stratification: Currently, PRS provide modest predictive value when used alone, as they primarily reflect inherited DNA variation rather than downstream biological states^[314,315]. Because PD risk is shaped not only by genetics but also by epigenetic regulation, immune-metabolic status, and environmental exposures, next-generation models are expected to integrate multiple layers - genome, epigenome, transcriptome, metabolome, and exposome - for more comprehensive risk assessment^[316]. For example, an individual carrying high-risk variants (e.g., LRRK2 or a high PRS) who also shows lipidomic changes such as elevated sphingolipids and pro-inflammatory cytokine profiles may have a substantially higher integrated risk estimate, whereas some high-risk gene carriers may remain unaffected when metabolic and immune indicators are unremarkable^[316-318]. Machine-learning methods can identify informative combinations across these modalities, but models require external validation and careful assessment of generalizability^[316]. In microbiome-based work, some single-cohort studies report high internal discrimination; however, a multi-study meta-analysis across thousands of samples found an average AUC of ~0.72, underscoring both promise and current limitations^[319]. Large longitudinal resources such as PPMI 2.0 are now assembling and sharing multimodal clinical, imaging, genetic, and biospecimen data to enable such integrative prediction efforts. Early multimodal-fusion studies using PPMI already show that combining imaging with clinical features improves discrimination of early/prodromal PD, illustrating a realistic near-term path for integrative models. As these tools move toward earlier-life risk estimation, ethical communication and counseling protocols for long-horizon risk prediction will be essential.

Artificial Intelligence Assisted Predictive Analytics: With the accumulation of multi-modal datasets, artificial intelligence (AI) is expected to play a pivotal role in early detection and disease trajectory modeling. Deep learning algorithms can extract features from complex genomic, microbiome, imaging, and behavioral data that may not be apparent through conventional analyses. Rather than directing treatment decisions, these tools primarily enable individualized risk estimation, progression forecasting, and dynamic monitoring, helping identify individuals who may benefit from closer clinical surveillance or additional testing. Such predictive models may facilitate earlier enrollment into observational studies or biomarker-validation cohorts, thereby improving understanding of disease evolution. Of course, achieving this vision requires collaboration among neurologists, data scientists, geneticists, and other interdisciplinary experts to jointly build credible models. Fortunately, several international teams have launched prototype projects to apply AI technology to early screening and comprehensive data analysis of PD.

In the future, PD diagnosis is expected to evolve toward early and more accurate detection-identifying subtle biological and behavioral changes years before motor symptoms emerge and providing quantitative risk profiles for individuals. This predictive framework aims to enable timely surveillance and improved staging rather than late recognition of established disease. Similar to cardiovascular risk scores or dementia prediction models, PD research is progressing toward scientifically grounded early risk estimation systems that support proactive clinical monitoring.