Radionucleotide imaging

Loss of dopaminergic cells in the substantia nigra, with resulting dopaminergic denervation of the striatum, is the hallmark of PD and is known to occur early in the pathologic process (Bernheimer et al., Reference Bernheimer, Birkmayer, Hornykiewicz, Jellinger and Seitelberger1973; Hughes et al., Reference Hughes, Daniel, Kilford and Lees1992). Imaging to capture the integrity of the striatal dopamine system thus has strong potential and biological plausibility in identifying prodromal PD. Various radionucleotide ligands have been applied. One of the most widely studied and available is an iodinated ligand of the dopamine transporter (DAT), a presynaptic membrane protein. DAT binding is decreased in PD due to degeneration of the presynaptic dopaminergic projections from the substantia nigra compacta (SNc) to the striatum.

Using single-photon emission computerized tomography (SPECT) imaging, radionucleotide ligand binding to DAT in PD patients was found to be decreased and correlated with severity of motor impairment (Huang et al., Reference Huang, Lin, Lin, Wey, Ting and Liu2001; Seibyl et al., Reference Seibyl, Marek, Quinlan, Sheff, Zoghbi, Zea-Ponce and van Dyck1995, Reference Seibyl, Marek, Sheff, Zoghbi, Baldwin, Charney and Innis1998). Cognitive impairment and behavioral symptoms (psychosis and depression) also correlated with decreased DAT SPECT binding in PD patients (Ravina et al., Reference Ravina, Marek, Eberly, Oakes, Kurlan, Ascherio and Shoulson2012). The sensitivity of DAT SPECT for PD diagnosis in a group of subjects with a parkinsonian syndrome is estimated to be 92% in comparison to a gold standard of movement disorder expert diagnosis at 6-month follow-up (Jennings et al., Reference Jennings, Seibyl, Oakes, Eberly, Murphy and Marek2004). However, while decreased DAT SPECT binding is highly specific for a neurodegenerative parkinsonian disorder, it does not reliably distinguish between the various neurodegenerative parkinsonian disorders [e.g., PD, multiple system atrophy (MSA), progressive supranuclear palsy (PSP), and dementia with Lewy bodies (DLB)].

Decreased DAT binding on imaging has been demonstrated in several clinically defined prodromal cohorts and longitudinal studies, suggesting that it indicates a high likelihood of developing PD motor symptoms. For example, in a cohort of RBD patients without a diagnosable neurodegenerative parkinsonian disorder at baseline, 40% (17 of 43) showed decreased striatal DAT binding (Iranzo et al., Reference Iranzo, Lomeña, Stockner, Valldeoriola, Vilaseca and Salamero2010). After an average follow-up of 21 months, 8 of 43 RBD patients developed a parkinsonian neurodegenerative disorder (PD or DLB), with 6 of these 8 having had decreased DAT binding at baseline testing (Iranzo et al., Reference Iranzo, Lomeña, Stockner, Valldeoriola, Vilaseca and Salamero2010). In another study, first-degree asymptomatic relatives of PD patients underwent smell testing and DAT SPECT imaging. Decreased DAT binding was demonstrated in a greater number of hyposmic versus normosmic individuals (Ponsen et al., Reference Ponsen, Stoffers, Wolters, Booij and Berendse2010). Among the 40 hyposmic individuals, 5 developed PD on 5-year follow-up; all 5 had decreased DAT binding on baseline imaging (Ponsen et al., Reference Ponsen, Stoffers, Wolters, Booij and Berendse2010).

In another cohort of individuals with hyposmia and first-degree relatives with PD, the PARS cohort, 11% of hyposmic individuals showed decreased DAT binding in comparison to 1% of normosmic individuals (Jennings et al., Reference Jennings, Siderowf, Stern, Seibyl, Eberly and Oakes2014). When additional clinical features, namely male sex and constipation, accompanied hyposmia in this cohort, 43% of this group showed a DAT deficit (Jennings et al., Reference Jennings, Siderowf, Stern, Seibyl, Eberly and Oakes2014). These studies highlight the utility of DAT imaging in identifying prodromal PD, particularly in combination with clinical data such as RBD or hyposmia.

PD-related changes in cerebral metabolism due to striatonigral denervation can also be imaged with fluorodopa (¹⁸F-DOPA) positron emission tomography (PET). Similar to studies performed with SPECT, ¹⁸F-DOPA PET imaging identifies decreased ¹⁸F-DOPA uptake in the caudate and putamen of PD patients. This has also been demonstrated in individuals with features of prodromal PD, namely RBD and depression, in comparison to healthy controls (Wing et al., Reference Wing, Lam, Zhang, Leung, Ho, Chen and Ho2015). Similarly, using the ligand [¹¹C]dihydrotetrabenazine, which binds to vesicular monoamine transporter 2, PET showed decreased striatal binding in a group of subjects with RBD in comparison to healthy controls (Albin et al., Reference Albin, Koeppe, Chervin, Consens, Wernette, Frey and Aldrich2000).

Preliminary studies of PET imaging in genetically defined prodromal individuals suggests this imaging modality may be of utility in this group as well. For example, in a study of seven asymptomatic parkin mutation carriers and seven PD patients with parkin mutation, ¹⁸F-DOPA PET was abnormal in three of the asymptomatic parkin carriers (compared to all of the manifesting parkin PD patients) (Walter et al., Reference Walter, Klein, Hilker, Benecke, Pramstaller and Dressler2004). Similarly, asymptomatic carriers of PINK1 mutations show decreased uptake of ¹⁸F-DOPA PET in comparison to controls (Khan et al., Reference Khan, Valente, Bentivoglio, Wood, Albanese, Brooks and Piccini2002). Longitudinal follow-up is needed to clarify the predictive value of baseline PET imaging, alone or in combination with other prodromal markers, in individuals genetically predisposed to PD, particularly among those with mutations that have incomplete penetrance.

¹⁸F-fluorodeoxyglucose (¹⁸FDG) PET imaging can be used to identify metabolic network patterns. In PD, ¹⁸FDG PET shows increased lentiform nucleus and thalamic metabolic activity, with decreased lateral frontal, paracentral, inferior parietal, and parieto-occipital activity. This pattern discriminates early stage symptomatic PD patients from healthy controls (Eidelberg et al., Reference Eidelberg, Moeller, Dhawan, Spetsieris, Takikawa, Ishikawa and Przedborski1994). In a study comparing individuals with RBD, PD, and healthy controls, the PD-associated metabolic network activity was increased in the RBD patients in comparison to the controls (Wu et al., Reference Wu, Yu, Peng, Dauvilliers, Wang, Ge and Zuo2014). In contrast, in a study of asymptomatic carriers of SNCA duplication (an established genetic cause of PD with incomplete penetrance) (Nishioka et al., Reference Nishioka, Hayashi, Farrer, Singleton, Yoshino, Imai and Hattori2006), no abnormalities in smell or changes in occipital lobe metabolism on ¹⁸FDG PET were found (Nishioka et al., Reference Nishioka, Ross, Ishii, Kachergus, Ishiwata, Kitagawa and Hattori2009), illustrating the importance of considering the reduced penetrance and phenotypic heterogeneity of mutations associated with PD, particularly in the prodromal phase.

Magnetic resonance imaging

While conventional clinical magnetic resonance imaging (MRI) sequences are not of utility in detecting prodromal PD, several more advanced MRI techniques show promise in this regard. A neuromelanin-sensitive MRI sequence on 3.0 Tesla MRI was designed to visualize neuromelanin-containing nuclei with greater detail. In PD, this imaging technique shows attenuation in the lateral SNc and locus coeruleus (LC) (Ohtsuka et al., Reference Ohtsuka, Sasaki, Konno, Koide, Kato, Takahashi and Terayama2013; Sasaki et al., Reference Sasaki, Shibata, Tohyama, Takahashi, Otsuka, Tsuchiya and Sakai2006). This attenuation discriminated early (median duration 1.5 years, Hoehn & Yahr stage 2) and late PD (median duration 12 years, Hoehn & Yahr stage 4) from healthy controls, but did not differ between the early and late PD groups (Ohtsuka et al., Reference Ohtsuka, Sasaki, Konno, Koide, Kato, Takahashi and Terayama2013). The sensitivity and specificity for discriminating early PD from healthy controls was 73% and 87% in the lateral SNc, and 82% and 90% in the LC (Ohtsuka et al., Reference Ohtsuka, Sasaki, Konno, Koide, Kato, Takahashi and Terayama2013). The higher sensitivity and specificity of attenuation in the LC compared to the SNc fits with Braak’s PD staging, in which LC involvement defines Braak stage 2, and SNc involvement defines Braak stage 3 (Braak et al., Reference Braak, Tredici, Rüb, de Vos, Jansen Steur and Braak2003). This modality has yet to be studied in prodromal PD, to our knowledge.

Another MRI modality, diffusion tensor imaging (DTI) has shown some potential in differentiating healthy controls from PD, with lower fractional anisotropy in the substantia nigra being the most consistent finding according to a recent meta-analysis (Cochrane & Ebmeier, Reference Cochrane and Ebmeier2013). In regards to application in prodromal cohorts, DTI in individuals with RBD in comparison to healthy controls shows changes in brainstem areas relevant to REM sleep, but not the substantia nigra, in a cross-sectional analysis (Scherfler et al., Reference Scherfler, Frauscher, Schocke, Iranzo, Gschliesser and Seppi2011). MRI with diffusion kurtosis imaging (DKI) can discriminate subjects with PD from healthy controls with higher sensitivity and specificity than DTI (Wang et al., Reference Wang, Lin, Lu, Weng, Ng, Wang and Wai2011). While promising, the utility of MRI in prodromal PD remains to be defined.

Ultrasound

Transcranial sonography (TCS) at the temporal bone windows allows for assessment of abnormal intracranial iron deposition. This is of relevance in PD as increased iron deposition is seen in the substantia nigra in PD compared to HC (Sofic et al., Reference Sofic, Riederer, Heinsen, Beckmann, Reynolds, Hebenstreit and Youdim1988), and is involved in PD pathophysiology (Faucheux et al., Reference Faucheux, Martin, Beaumont, Hauw, Agid and Hirsch2003). Hyperechogenicity in the substantia nigra correlates with higher iron levels on postmortem studies (Berg et al., Reference Berg, Roggendorf, Schröder, Klein, Tatschner, Benz and Becker2002). The prevalence of substantia nigra hyperechogenicity in PD patients is approximately 90% (Berg, Siefker, & Becker, Reference Berg, Siefker and Becker2001), compared to 9–19% (Berg et al., Reference Berg, Becker, Zeiler, Tucha, Hofmann, Preier and Lange1999, Reference Berg, Godau, Seppi, Behnke, Liepelt-Scarfone and Lerche2013) in community-dwelling older adults without PD.

TCS has been applied in both genetic and clinical prodromal cohorts and shows great promise as an imaging biomarker for the prodromal PD state. In a study of TCS in subjects with PD and healthy controls, the subset of controls with substantia nigra hyperechogenicity were also found to have decreased dopamine binding on ⁸F-DOPA PET, suggesting they have increased risk of developing PD symptoms in the future (Berg et al., Reference Berg, Becker, Zeiler, Tucha, Hofmann, Preier and Lange1999). Substantia nigra hyperechogenicity on TCS in a community sample of older adults without PD has been found to have a sensitivity and specificity of 80% and 81% for development of PD over 3 years (Berg et al., Reference Berg, Godau, Seppi, Behnke, Liepelt-Scarfone and Lerche2013). In this same cohort, if both hyposmia and a family history of PD are present, sensitivity and specificity of substantia nigra hyperechogenicity for development of PD over 3 years increases to 80% and 91% (Berg et al., Reference Berg, Godau, Seppi, Behnke, Liepelt-Scarfone and Lerche2013).

In a study of RBD patients without parkinsonism, substantia nigra hyperechogenicity was found in 36% (14 of 30) of individuals with RBD, compared to 11% (16/149) of healthy controls (Iranzo et al., Reference Iranzo, Lomeña, Stockner, Valldeoriola, Vilaseca and Salamero2010). Five of the 14 individuals with RBD and substantia nigra hyperechogenicity went on to develop parkinsonism at 21 months follow-up (Iranzo et al., Reference Iranzo, Lomeña, Stockner, Valldeoriola, Vilaseca and Salamero2010). This study found combining substantia nigra hyperechogenicity on TCS and decreased DAT SPECT binding yielded a combined sensitivity of 100% and specificity 55% in identifying individuals with RBD who later developed parkinsonism (Iranzo et al., Reference Iranzo, Lomeña, Stockner, Valldeoriola, Vilaseca and Salamero2010).

The utility of combining easily obtained, low cost biomarkers with more specific and yet more logistically demanding ones was demonstrated in a population-based study that incorporated TCS. Four groups were identified: (i) idiopathic PD, (ii) presence of parkinsonian signs possibly due to a neurodegenerative parkinsonism, (iii) presence of non-specific motor abnormalities presumed to be due to non-neurologic etiologies (e.g., arthritis), and (iv) healthy controls (Tunc et al., Reference Tunc, Graf, Tadic, Brüggemann, Schmidt, Al-Khaled and Kasten2015). Hyperechogenicity in the substantia nigra on TCS had a sensitivity of 76.6% and a specificity of 86.5% for PD diagnosis (Tunc et al., Reference Tunc, Graf, Tadic, Brüggemann, Schmidt, Al-Khaled and Kasten2015). In comparison, hyposmia had a sensitivity of 68.1% and a specificity of 74.9% for PD diagnosis (Tunc et al., Reference Tunc, Graf, Tadic, Brüggemann, Schmidt, Al-Khaled and Kasten2015). The combination of both hyperechogenicity in the substantia nigra and hyposmia yielded an excellent specificity of 97.7%, but sensitivity was reduced to 51.1% (Tunc et al., Reference Tunc, Graf, Tadic, Brüggemann, Schmidt, Al-Khaled and Kasten2015).

With regard to application in cohorts with incompletely penetrant genetic mutations that cause PD, substantia nigra hyperechogenicity is not different between idiopathic PD and G2019S LRRK2 PD patients; asymptomatic LRRK2 carriers have less substantia nigra hyperechogenicity compared to PD patients, but more substantia nigra hyperechogenicity than controls (Brüggemann et al., 2011). Similarly, in a study of seven asymptomatic parkin mutation carriers and seven PD patients with parkin mutation, substantia nigra hyperechogenicity was found in all PD patients with parkin mutation and five parkin carriers (Walter et al., Reference Walter, Klein, Hilker, Benecke, Pramstaller and Dressler2004).

Cardiac scintography

Cardiac ¹²³I-metaiodobenzylguanidine (MIBG) scintigram is a nuclear imaging technique used to assess postganglionic sympathetic cardiac innervation. In PD, cardiac denervation due to vagal nerve nucleus involvement is expected to occur early, at Braak’s stage 1 (Braak et al., Reference Braak, Tredici, Rüb, de Vos, Jansen Steur and Braak2003). The primary assessment of a cardiac scintigram is the ratio of heart-to mediastinum (H/M) MIBG accumulation at early (20 min) and late (4 hr) intervals. In PD, the sensitivity and specificity for early MIBG scintigram ratios is 81% and 85%; for late ratios it is 84% and 90% (Sawada et al., Reference Sawada, Oeda, Yamamoto, Kitagawa, Mizuta, Hosokawa and Kawamura2009). In early PD, defined in this study as disease duration less than 3 years, sensitivity decreases to 76% and 74%, respectively, for early and late H/M accumulation (Sawada et al., Reference Sawada, Oeda, Yamamoto, Kitagawa, Mizuta, Hosokawa and Kawamura2009).

The utility of cardiac MIBG scintography in prodromal PD is not fully defined. MIBG uptake was reduced in a group of individuals with RBD compared to controls (Kashihara, Imamura, & Shinya, Reference Kashihara, Imamura and Shinya2010; Miyamoto et al., Reference Miyamoto, Miyamoto, Inoue, Usui, Suzuki and Hirata2006). Application of MIBG in other prodromal groups has not been reported, to our knowledge. Of note, in patients with PD due to known genetic causes (parkin, DJ-1, PINK1, and G2019S LRRK2 gene mutation), cardiac MIBG scintigram often shows preserved MIBG accumulation ratios (Orimo et al., Reference Orimo, Amino, Yokochi, Kojo, Uchihara, Takahashi and Mizuno2005; Quattrone et al., Reference Quattrone, Bagnato, Annesi, Novellino, Morgante, Savettieri and Condino2008). If similar results are seen in asymptomatic carriers of these mutations, this would suggest that cardiac MIBG scintigram may not be useful in genetically defined prodromal PD.

Blood

Given that abnormal alpha-synuclein accumulation is a primary component of PD pathology, body fluid alpha-synuclein has long been sought after as a potential biomarker candidate. In the blood, the ratio of red blood cell (RBC) oligomeric/total alpha-synuclein is higher in PD patients than controls with a sensitivity of 79% and specificity of 65% (Wang, Yu, Li, & Feng, Reference Wang, Yu, Li and Feng2015). Elevated plasma alpha-synuclein oligomer levels have been found in PD patients compared to controls with a sensitivity of 53% and specificity 86% (El-Agnaf et al., Reference El-Agnaf, Salem, Paleologou, Curran, Gibson, Court and Allsop2006). However, the same group was later unable to replicate these results for total or oligomeric alpha-synuclein, normal or phosphorylated (Foulds et al., Reference Foulds, Mitchell, Parker, Turner, Green, Diggle and Allsop2011). It is apparent from their work and others that the strain of alpha-synuclein measured and the techniques used largely influence the results of such studies (Malek et al., Reference Malek, Swallow, Grosset, Anichtchik, Spillantini and Grosset2014).

DJ-1 is a protein related to oxidative stress and was initially linked to PD after familial cases of PD were found to be caused by a mutation in the gene encoding DJ-1. A study of PD and DLB patients showed elevated DJ-1 plasma levels compared to healthy controls (Waragai et al., Reference Waragai, Nakai, Wei, Fujita, Mizuno, Ho and Hashimoto2007). This was contradicted in a subsequent study by a different group who looked at DJ-1 levels in platelet-free plasma, comparing PD patients to a group of healthy and Alzheimer’s disease (AD) controls (Shi et al., Reference Shi, Zabetian, Hancock, Ginghina, Hong, Yearout and Zhang2010). Yet, a later study by the same group looked at DJ-1 isoforms in whole blood samples and found differences between these groups, with some of the differences only demonstrable in late-stage PD (Lin et al., Reference Lin, Cook, Zabetian, Leverenz, Peskind, Hu and Zhang2012). Thus, similar to alpha-synuclein, serum DJ-1 is a potential biomarker of the prodromal PD state, but much remains to be learned before it can be applied.

Serum uric acid, an antioxidant, has consistently been found to be lower in PD patients compared to controls, and is inversely correlated with both PD risk and PD disease progression (Ascherio et al., Reference Ascherio, LeWitt, Xu, Eberly, Watts and Matson2009; Davis et al., Reference Davis, Grandinetti, Waslien, Ross, White and Morens1996; de Lau, Koudstaal, Hofman, & Breteler, Reference de Lau, Koudstaal, Hofman and Breteler2005; Schwarzschild et al., Reference Schwarzschild, Schwid, Marek, Watts, Lang, Oakes and Ondrasik2008; Weisskopf, O’Reilly, Chen, Schwarzschild, & Ascherio, Reference Weisskopf, O’Reilly, Chen, Schwarzschild and Ascherio2007). Uric acid levels are also inversely correlated with the likelihood of non-motor symptoms in PD (Moccia et al., Reference Moccia, Picillo, Erro, Vitale, Longo, Amboni and Pellecchia2014, Reference Moccia, Picillo, Erro, Vitale, Longo, Amboni and Pellecchia2015). This has been one of the most consistent and replicated findings in regard to potential PD biomarkers. Preliminary studies show this may extend to the prodromal PD state. A metabolomics profiling study of asymptomatic carriers of the G2019S LRRK2 mutation showed that they had significantly lower uric acid levels compared to controls (Johansen et al., Reference Johansen, Wang, Aasly, White, Matson, Henchcliffe and Bogdanov2009).

Apolipoprotein A1, a serum protein, is a major component of high-density lipoprotein (HDL) and functions in lipid metabolism. Decreased plasma apolipoprotein A1 levels correlate with increased PD risk and decreased age of PD onset, a finding that was subsequently replicated in an independent cohort (Qiang et al., Reference Qiang, Wong, Siderowf, Hurtig, Xie, Lee and Chen-Plotkin2013). Importantly, in a cohort of asymptomatic people at high-risk of PD (the PARS cohort), a group enriched for hyposmia, low plasma apolipoprotein A1 was associated with decreased binding on DAT SPECT (Qiang et al., Reference Qiang, Wong, Siderowf, Hurtig, Xie, Lee and Chen-Plotkin2013). This makes apolipoprotein A1 a prime candidate for a serum biomarker in prodromal PD.

Several inflammatory markers have been proposed as biomarkers for PD. A few studies have found higher levels of interleukin-6 (IL-6) in PD patients compared to controls (Brodacki et al., Reference Brodacki, Staszewski, Toczyłowska, Kozłowska, Drela, Chalimoniuk and Stepien2008; Chen, O’Reilly, Schwarzschild, & Ascherio, Reference Chen, O’Reilly, Schwarzschild and Ascherio2008), although one study found lower IL-6 levels (Dursun et al., Reference Dursun, Gezen-Ak, Hanağası, Bilgiç, Lohmann, Ertan and Yılmazer2015). In a study of PD patients with GBA mutations, IL-8 was found to be elevated in a discovery and replication cohort, but not elevated in idiopathic PD patients without GBA mutation (Chahine et al., Reference Chahine, Qiang, Ashbridge, Minger, Yearout, Horn and Chen-Plotkin2013).

In addition, metabolomics have been applied to identify potential serum PD biomarkers. A metabolomics study of 43 drug-naïve PD patients and 37 healthy controls showed increased plasma pyruvate (Ahmed, Santosh, Kumar, & Christlet, Reference Ahmed, Santosh, Kumar and Christlet2009). A different metabolomics study showed uric acid is decreased and glutathione is increased in a sample of 66 PD patients compared to 25 controls (Bogdanov et al., Reference Bogdanov, Matson, Wang, Matson, Saunders-Pullman, Bressman and Beal2008).

Finally, differential gene expression patterns in PD have been pursued as a biomarker. Using a microarray on whole blood samples to look for a transcriptional mRNA PD signature, one group found a panel of expression of eight genes was associated with PD in comparison to a control group of healthy, AD, and PSP patients (Scherzer et al., Reference Scherzer, Eklund, Morse, Liao, Locascio, Fefer and Gullans2007). This gene expression pattern was validated in a second PD sample; its positivity confers an odds of PDF of 5.1 (Scherzer et al., Reference Scherzer, Eklund, Morse, Liao, Locascio, Fefer and Gullans2007). In a cross-sectional study of mRNA expression profiles in an Ashkenazi Jewish cohort of asymptomatic individuals and PD patients, mRNA expression pattern distinguished both PD disease state and LRRK2 genotype (Chikina et al., Reference Chikina, Gerald, Li, Ge, Pincas, Nair and Sealfon2015).

Inflammatory markers, metabolomics, and gene expression studies have not yet been applied widely to prodromal cohorts. As serum biomarkers for PD are identified, their validation and replication in independent cohorts will be essential, as will be studies of their utility in prodromal PD.

Cerebrospinal fluid

As in the case of the search for potential PD biomarkers in the blood, the most studied biomarker candidate in the cerebrospinal fluid (CSF) is alpha-synuclein. Studies of CSF alpha-synuclein and alpha-synuclein oligomers have shown mixed results, possibly due to heterogeneity in collection and testing techniques, as well as the potential for contamination with blood, which can falsely elevate measured CSF alpha-synuclein levels (Malek et al., Reference Malek, Swallow, Grosset, Anichtchik, Spillantini and Grosset2014). Biological factors may also play a role in the variability of the results, which were reviewed in depth by Mollenhauer et al. (2015). A recent meta-analysis of 12 studies found CSF alpha-synuclein levels significantly differed between PD, MSA, and controls, with no difference between PD and DLB or PSP (Zhou, Wen, Yu, Zhang, & Jiao, Reference Zhou, Wen, Yu, Zhang and Jiao2015). A different meta-analysis, of 17 studies, found no difference in CSF alpha-synuclein levels between PD, DLB, and MSA patients, yielding an estimated sensitivity and specificity of 88% and 40% (Gao et al., Reference Gao, Tang, Nie, Wang, Zhao, Gan and Wang2015). Given the role of alpha-synuclein in PD pathophysiology and the goal of targeting this protein to prevent PD, further CSF studies of alpha-synuclein are eagerly awaited.

As in the blood, DJ-1 protein is also present in CSF. A study comparing a group of PD patients to a control group of healthy individuals and AD patients found a sensitivity and specificity of DJ-1 in the CSF of 90% and 70%, respectively, for PD (Hong et al., Reference Hong, Shi, Chung, Quinn, Peskind, Galasko and Zhang2010). In a study of PD, MSA, and healthy controls, DJ-1 levels were found to be highest in MSA, followed by PD, and then controls, discriminating MSA from PD with a sensitivity of 78% and specificity of 78%; and discriminating PD from controls with a sensitivity of 81% and specificity of 52% (Herbert et al., Reference Herbert, Eeftens, Aerts, Esselink, Bloem, Kuiperij and Verbeek2014).

Studies on the utility of AD biomarkers in PD have yielded mixed results (Jiménez-Jiménez, Alonso-Navarro, García-Martín, & Agúndez, Reference Jiménez-Jiménez, Alonso-Navarro, García-Martín and Agúndez2014). Attempts have been made to use CSF proteins in combination to improve diagnostic accuracy. For example, in a study comparing healthy controls to early PD patients (0.4 year median disease duration), only lower beta-amyloid 1–42 and phosphorylated tau levels were associated with PD diagnosis on multivariate regression, although all tested CSF biomarkers, including total tau, alpha-synuclein, and total tau/beta-amyloid 1–42 ratio, were slightly lower in PD patients compared to controls (Kang et al., Reference Kang, Irwin, Chen-Plotkin, Siderowf, Caspell and Coffey2013). Studies of these putative CSF biomarkers in prodromal PD cohorts have not been published to our knowledge.

Several other CSF biomarkers have been investigated as PD biomarkers and show promise. However, much remains to be learned about them and their specificity for PD versus their utility more as non-specific markers of neurodegeneration. For example, neurofilament light chain (NFL) levels are elevated in the CSF in some neurodegenerative diseases, but not in PD in comparison to healthy controls (Constantinescu, Rosengren, Johnels, Zetterberg, & Holmberg, Reference Constantinescu, Rosengren, Johnels, Zetterberg and Holmberg2010; Hall et al., Reference Hall, Öhrfelt, Constantinescu, Andreasson, Surova, Bostrom and Hansson2012). A meta-analysis of 4 studies showed elevated NFL levels in MSA and PSP compared to PD (Sako, Murakami, Izumi, & Kaji, Reference Sako, Murakami, Izumi and Kaji2015). Similarly, cell-free circulating mitochondrial DNA (ccf-mtDNA) is reduced in CSF of PD patients compared to controls, but it is also decreased in AD (Pyle et al., Reference Pyle, Brennan, Kurzawa-Akanbi, Yarnall, Thouin, Mollenhauer and Hudson2015).

Nuclear factor (erythroid-derived 2)-like 2 (Nrf2), a CSF protein, did not discriminate between PD and controls, but Nrf2 concentrations did correlate with motor scores in 1 study of LRRK2 positive PD subjects (Loeffler, Smith, Coffey, Aasly, & LeWitt, Reference Loeffler, Smith, Coffey, Aasly and LeWitt2015).

There are inconsistent findings on the utility of dopamine and serotonin metabolites in discriminating PD from healthy controls, and for their potential for correlation with PD features (Jiménez-Jiménez et al., Reference Jiménez-Jiménez, Alonso-Navarro, García-Martín and Agúndez2014). To our knowledge, these have not been studied in prodromal PD cohorts.

As with serum metabolomic studies, CSF metabolomics is a promising area as well. Nuclear magnetic resonance metabolomics uses spectroscopy to quantify metabolites in biofluids. A study in 10 PD patients and 10 healthy controls found lower CSF alanine, creatinine, and mannose levels in the PD patients sufficient to discriminate between the groups (Öhman & Forsgren, Reference Öhman and Forsgren2015). Further validation and replication in PD and prodromal PD is required.

Skin, colon, salivary glands

Multiple studies have found alpha-synuclein in several types of peripheral tissues (Malek et al., Reference Malek, Swallow, Grosset, Anichtchik, Spillantini and Grosset2014). In 2006, Braak, de Vos, and Del Tredici reported alpha-synuclein in the enteric nervous system in post mortem examination of PD patients, which they proposed could reflect the entry point of the pathology to the vagus nerve, stage 1 of the Braak PD pathology staging system (Braak et al., Reference Braak, Tredici, Rüb, de Vos, Jansen Steur and Braak2003). Multiple studies have since reported increased colonic alpha-synuclein expression in PD and pre-motor PD in comparison to controls (Gold, Turkalp, & Munoz, Reference Gold, Turkalp and Munoz2013; Lebouvier et al., Reference Lebouvier, Chaumette, Damier, Coron, Touchefeu, Vrignaud and Neunlist2008; Pouclet et al., Reference Pouclet, Lebouvier, Coron, Des Varannes, Neunlist and Derkinderen2012; Shannon, Keshavarzian, Dodiya, Jakate, & Kordower, Reference Shannon, Keshavarzian, Dodiya, Jakate and Kordower2012).

In one study, colonic biopsies sampled from patients 2–5 years before the development of PD motor symptoms were found to exhibit alpha-synuclein pathology (Shannon et al., Reference Shannon, Keshavarzian, Dodiya, Jakate and Kordower2012). However, the possibility that colonic alpha-synuclein can discriminate PD and prodromal PD from controls came into question when a study of 22 PD patients and 11 controls found alpha-synuclein in the colon of all patients and controls (Visanji et al., Reference Visanji, Marras, Kern, Al Dakheel, Gao, Liu and Hazrati2015). The differing results from studies of colonic alpha-synuclein may be partially due to differences in collection techniques and staining (Visanji, Marras, Hazrati, Liu, & Lang, Reference Visanji, Marras, Hazrati, Liu and Lang2014), as demonstrated in a study comparing colonic alpha-synuclein levels in RBD, PD, and controls. It ultimately failed to show any significant differences between the groups, and provided further evidence that differences in biopsy depth, location, and staining technique affect results (Sprenger et al., Reference Sprenger, Stefanova, Gelpi, Seppi, Navarro-Otano, Offner and Poewe2015). Further studies of colonic alpha-synuclein with refinement and standardization of techniques should continue as it could hold great potential as a prodromal PD biomarker.

In the skin, alpha-synuclein has been found in idiopathic PD patients, but not in patients with other types of parkinsonism or controls (Donadio et al., Reference Donadio, Incensi, Leta, Giannoccaro, Scaglione, Martinelli and Liguori2014; Rodríguez-Leyva et al., Reference Rodríguez-Leyva, Calderón-Garcidueñas, Jiménez-Capdeville, Rentería-Palomo, Hernandez-Rodriguez, Valdés-Rodríguez and Castanedo-Cázares2014). In addition, salivary gland alpha-synuclein has been found to have high specificity for PD in most studies (Beach et al., Reference Beach, Adler, Dugger, Serrano, Hidalgo and Henry-Watson2013; Tredici, Hawkes, Ghebremedhin, & Braak, Reference Tredici, Hawkes, Ghebremedhin and Braak2010), but not all (Folgoas et al., Reference Folgoas, Lebouvier, Leclair-Visonneau, Cersosimo, Barthelaix, Derkinderen and Letournel2013).

Alpha-synuclein testing in peripheral tissues holds great promise for potential PD biomarkers. Once sampling and testing procedures have been refined, application to prodromal PD cohorts will be of great interest.