Background
Parkinson disease (PD), characterized by its motor symptoms (bradykinesia, resting tremor, and rigidity) [
1], does not start suddenly. By the time the clinical diagnosis has been made, some 30–50% of dopaminergic neurons have been lost in the substantia nigra [
2]. Symptomatic treatments are effective in most patients with PD, but currently no drugs have demonstrated convincing evidence of disease modification. One possible explanation is that the pathology of PD may be sufficiently advanced at the point of diagnosis that none of the interventions can rescue the remaining dying neurons, thus the prodromal stage of PD, during which the disease pathology has started but is insufficient to result in clinical manifestations, provides a valuable window during which disease-modifying therapies can be tested [
3].
According to recent Movement Disorder Society criteria, early PD can be divided into three stages: preclinical PD (neurodegeneration has started yet without evident symptoms and signs); prodromal PD (symptoms and signs are present, but are still insufficient to define PD) and clinical PD (diagnosis of PD based on classical symptoms). The criteria are based upon probability and likelihood since it is not possible to identify prodromal PD with 100% certainty; probable prodromal PD is defined as a high likelihood (greater than 80%) and possible prodromal PD as a likelihood between 30 and 80% [
4,
5]. The cardinal features of prodromal PD are non-motor and include constipation, hyposmia/anosmia, depression, REM sleep behavior disorder, orthostatic hypotension, and loss of heart rate variability [
6]. Notably, many of the symptoms that emerge earlier in the disease course can be attributed to dysfunction in the peripheral nervous system or the peripheral part of the central nervous system, such as the vagus nerve (e.g. constipation), the sympathetic nervous system (e.g. orthostatic hypotension), or the olfactory bulb (hyposmia).
Neuronal aggregation of alpha-synuclein (α-syn) in Lewy bodies and Lewy neurites, the pathological signature of sporadic PD, can be found in the peripheral nervous system of PD patients [
7]. It is not clear whether these structures are the original site of α-syn aggregation or whether they are subject to α-syn pathology transported from the brain. In support of the former hypothesis, truncal vagotomy has been associated with a reduced risk of PD after 20 years of follow-up (adjusted hazard ratio [HR] = 0.53; 95% CI: 0.28–0.99) [
8]. Based on evidence from human studies, cell culture and animal models, the paradigm of pathological protein propagation in neurodegenerative diseases has been extended to include the concept that pathology arising from neurodegeneration-related proteins such as α-syn, amyloid-β, tau and TAR DNA-binding protein 43 (TDP43) may propagate in a prion-like fashion [
9‐
13]. On the other hand, the prion hypothesis as selective neuronal vulnerability may be another important factor contributing to specific patterns of degeneration in human and animal brains [
13]. In PD patients who underwent human fetal nigral transplantation, Lewy body-like inclusions that stained positive for α-syn were found in the grafted nigral neurons 14 years after transplantation, suggestive of cell to cell transmission [
14,
15]. It is hypothesized that the propagation of α-syn in the brain starts in the dorsal motor nucleus of the glossopharyngeal and vagus nerves (DMV) and the olfactory bulb; from these two structures the α-syn pathology spreads in an ascending pattern to the pons, the midbrain, the basal forebrain and finally to the neocortex through chains of vulnerable neurons [
16‐
18]. The so-called “Braak hypothesis” provides a mechanistic underpinning for the prodromal stage of PD, as non-motor symptoms could be explained by pathology in the peripheral nervous system and caudal brainstem that precede the onset of classic motor symptoms which do not emerge until Lewy pathology affects the substantia nigra. In this review we consider the two potential portals through which abnormal α-syn can access the brain: the vagus nerve and the olfactory bulb. We review clinical, pathological and neuroimaging evidence, and suggest future directions for studies in prodromal disease.
The internal homogeneity and heterogeneity of prodromal mechanisms
In fact, the linkage between different prodromal symptoms and imaging signs of prodromal PD are universal. Hyposmia has been associated with constipation, depression, anxiety and mild motor symptoms [
45], a combination of symptoms is more predictive of decreased DAT binding [
22]. Other studies showed linkage between hyposmia, symptoms of autonomic failure and imaging evidence of sympathetic system denervation, such as lower cardiac septal: hepatic ratios of 6-
18F-fluorodopamine-derived radioactivity and lower cardiac
123I-metaiodobenzylguanidine uptake [
84‐
86]. In both manifest PD with RBD and idiopathic RBD patients, RBD has been linked with hyposmia, constipation, orthostatic symptoms, hallucinations, depression and worse parkinsonian sign [
87,
88]. In population-based studies, substantia nigra hyperechogenicity has been associated with constipation, hyposmia, depression and mild parkinsonian signs [
89].
The cause of this clustering of motor and non-motor symptoms is unknown, although different classifications of empirical subtypes based on the clusters are proposed [
90], the phenomena may simply follow the severity of pathological development of PD. Hyposmia, RBD and constipation constantly appear in different clusters, while the corresponding pathological structures are either the potential portals for α-syn aggregation (DMV and olfactory system) or are close to them (locus coeruleus/subcoeruleus complex and pedunculopotine nucleus), so it is natural that the symptoms should cluster together if α-syn propagates though the relevant structures. In support of this view, some evidence showed possible higher α-syn burden in subjects with hyposmia, RBD and reduced
123I-metaiodobenzylguanidine uptake [
91‐
93], in agreement with the Braak stage and the progression of PD. From this perspective, the homogeneity in the development of parkinsonian pathology is emphasized, and the recently described research criteria for prodromal PD assign each symptom and sign in those clusters into a combined score to predict future PD manifestation [
5].
On the other hand, such a scheme may neglect important heterogeneity of mechanisms in the development of PD. Braak and colleagues have proposed a dual-hit hypothesis in which a neurotropic pathogen might enter the brain through either the gastrointestinal or the nasal route [
94], either of which can result in disease progression, but potentially with different manifestations [
95,
96]. Empirical nonmotor subtypes are recently proposed, which categorize patients into brainstem phenotype (brainstem route, characterized with late onset hyposmia, RBD and dysautonomia), limbic phenotype (olfactory route, characterized by anosmia, depression, fatigue and central pain) and cognitive phenotype (diffused, characterized by cognitive decline) [
97,
98]. So far, no pathological evidence is available to support such subtyping and the internal axonal linkage between the olfactory bulb, olfactory cortex and basal forebrain, hypothalamus, and brainstem may introduce ambiguity in the separation of the two hypothetical routes [
99,
100]. However, functional and structural network analysis based on neuroimaging may help to investigate the real propagation patterns of α-syn pathology in the brain.
Another illustration of heterogeneity in PD is based on genetic subtypes, as there is evidence of pathophysiological differences related to certain gene mutations, such as increased inflammation in LRRK2 mutation carriers [
101,
102]. The lack or lesser extent of α-syn deposition in some genetic forms of PD further emphasizes these differences [
103]. Compared to RBD patients, LRRK2 carriers have significantly lower prevalence of olfactory loss, cognitive decline or sleep disturbance in the prodromal stage [
104‐
108]. Neuroimaging studies are needed to consider the functional and structural network changes in the genetic subtypes and to evaluate the differences between the sporadic subtypes and genetic subtypes in both non-manifest and manifest stages.
Even though not emphasized in this review, the sympathetic nervous system may deserve more attention in attempting to understand mechanisms of prodromal PD, as there is evidences for pre-motor involvement of peripheral noradrenergic depletion [
109], while the noradrenergic nucleus locus coeruleus may be affected prior to the substantia nigra in the prodromal stage. Related biomarker such as
123I-metaiodobenzylguanidine uptake and 3-methoxy-4-hydroxyphenylglycol can be potential early indicators for central neurodegeneration [
110].