Two- Year observational study of autonomic skin function in patients with Parkinson's disease compared to healthy individuals

Background and purpose: We characterized autonomic pilomotor and sudomotor skin function in early Parkinson's disease (PD) longitudinally. Methods: We enrolled PD patients (Hoehn and Yahr 1– 2) and healthy controls from movement disorder centers in Germany, Hungary, and the United States. We evaluated axon- reflex responses in adrenergic sympathetic pilomotor nerves and in cholinergic sudomotor nerves and assessed sympathetic skin response (SSR), predominantly parasympathetic neurocardiac function via heart rate variability, and disease- related symptoms at baseline, after 2 weeks, and after 1 and 2 years. Clini caltr ials.gov: NCT03043768. Results: We included 38 participants: 26 PD (60% females, aged 62.4 ± 7.4 years, mean ± SD) and 12 controls (75% females, aged 59.5 ± 5.8 years). Pilomotor function was reduced in PD compared to controls at baseline when quantified via spatial axon- reflex spread (78 [43– 143], median [interquartile range] mm 2 vs. 175 [68– 200] mm 2 , p = 0.01)


INTRODUC TI ON
Early diagnosis enables timely symptomatic treatment and improvement of quality of life in patients with Parkinson's disease (PD) [1]. However, detection of prodromal and early disease stages is challenging, as few diagnostic techniques are available [2]. The autonomic nervous system poses a promising diagnostic target, as dysautonomia in PD patients, including impaired thermoregulation and cardiovascular symptoms, often precedes onset of motor symptoms [3]. Although the pathophysiology of autonomic nervous system involvement in PD is not fully understood, accumulating evidence suggests that in early disease stages alpha-synuclein depositions damage small autonomic skin nerve fibers [4,5]. In PD patients, structural changes of cutaneous cholinergic sudomotor fibers and adrenergic pilomotor fibers correlate with severity of autonomic dysfunction and motor symptoms [6]. Clinical assessment of these structural changes via immunohistochemical analysis of skin biopsies is limited by its invasive nature. In previous research, we showed that integrity of pilomotor and sudomotor nerve fibers can be assessed noninvasively by utilizing the axon-reflex [7,8]. The axon-reflex is a neurogenic response, which is induced by cutaneous iontophoresis of either phenylephrine for adrenergic nerve fibers or acetylcholine for cholinergic fibers. Stimulation of adrenergic pilomotor nerve fibers leads to pilomotor erection (goose bumps), whereas activation of cholinergic sudomotor fibers results in local sweating. Quantitative analysis of these responses with spatial resolution using quantitative pilomotor axon-reflex test (QPART) and the quantitative direct and indirect test of sudomotor function (QDIRT) provides a measure of autonomic small nerve fiber integrity [7][8][9].
We demonstrated previously that functional integrity of pilomotor nerve fibers assessed via QPART is impaired in early stages of PD, indicating the potential value of this diagnostic target [10].
Here, we aimed to assess progression of small fiber dysfunction in PD over time and assess the utility of axon-reflex-based assessment of adrenergic and cholinergic small fiber integrity to detect and monitor peripheral autonomic impairment in these patients. Moreover, we sought to assess the reproducibility of QPART and QDIRT in PD patients and confirm the capability of both techniques to differentiate patients in early stages of PD from healthy subjects. Lastly, we aimed to evaluate consistency of observations and characterize topography of cutaneous small fiber function by repeating assessment on all extremities bilaterally.

Participants and protocol
We recruited male and female patients with mild PD (Hoehn and Yahr score 1-2) as well as healthy control subjects from university hospital-based PD outpatient clinics at four study sites (University Hospital Carl Gustav Carus, Dresden, Germany; Charité University Medicine Berlin, Berlin, Germany; Beth Israel Deaconess Medical Center, Boston, MA, United States; Semmelweis University, Budapest, Hungary).
Age range was set to 35-80 years. Detailed patient history was obtained and thorough neurological examination was performed at baseline to rule out clinically apparent large or small fiber neuropathy.
Between-group differences increased over the course of 2 years (p < 0.05), although no decline was observed within groups (p = ns). Pilomotor impairment in PD correlated with motor symptoms (rho = −0.59, p = 0.017) and was not lateralized (p = ns). Sudomotor axon-reflex and neurocardiac function did not differ between groups (p = ns), but SSR was reduced in PD (p = 0.0001).
Conclusions: Impairment of adrenergic sympathetic pilomotor function and SSR in evolving PD is not paralleled by changes to cholinergic sudomotor function and parasympathetic neurocardiac function, suggesting a sympathetic pathophysiology. A pilomotor axon-reflex test might be useful to monitor PD-related pathology.

K E Y W O R D S
autonomic nervous system, nerve, Parkinson disease, skin, synucleinopathy autonomic failure, inflammatory demyelinating polyradiculoneuropathies, multiple system atrophy, atypical Parkinson syndromes, body mass index > 25 kg/m 2 , acute or chronic renal disease, gout, rheumatoid arthritis, lupus, Sjögren syndrome, triple-A syndrome, and autonomic neuropathies not related to PD. Additional exclusion criteria for healthy control subjects included any acute or chronic disease and chronic intake of medication. All subjects were asked to avoid caffeine on the days of testing. Patients were evaluated at baseline, after 2 weeks (pilomotor only), after 1 year, and after 2 years. Study timeline and skin testing sites are depicted in Figure 1.

Experimental setup and testing environment
All assessments were performed in temperature-controlled rooms at 18-22°C at all participating sites. A detailed description of the experimental setup and environment is provided in Text S1.1.

Quantitative pilomotor axon-reflex test
The QPART had been previously introduced by our group as a test of adrenergic sympathetic skin nerve fibers and was further refined in this study [7]. In brief, we performed iontophoresis with 0.5 mA over 5 min to deliver 0.01% phenylephrine solution to the superficial skin layer by a drug delivery capsule electrode (LI-611, Perimed, Jakobsberg, Sweden) affixed on the cutaneous testing area. Resulting pilomotor erection in the "indirect" area of axon-reflex-mediated response adjacent to the "direct" iontophoresis stimulation area was then captured by silicone imprints to quantify the axon-reflex response. We built a semiautomated analysis algorithm using the open source tool ImageJ (National Institutes of Health, Bethesda, MD, United States) to improve the previous analysis protocol. Number of indents of erect hair follicles in the indirect area and spatial axonreflex spread were computed. The procedure is detailed in Text S1.1 and depicted in Figure 2.

Quantitative direct and indirect test of sudomotor function
We quantified postganglionic axon-reflex-mediated sweating in response to stimulation of sympathetic cholinergic skin nerve fibers using iontophoresis of 10% acetylcholine on the volar aspect of the forearm. The local sweat response was made visible using an indicator dye (povidone-iodine and cornstarch) followed by repeated digital photographs taken every 15 s for 8 min. Sweat droplets in the "indirect" axon-reflex region were quantified by number and the axon-reflex spread area was calculated as previously described [8]. We modified the analysis protocol by developing a semiautomated segmentation algorithm using an open source tool (ImageJ). The procedure is detailed in Text S1.2 and illustrated in Figure S1.
F I G U R E 1 Study timeline and testing sites. Illustration shows the time points of measurement (timeline) and skin sites of assessment. HRV, heart rate variability; QDIRT, quantitative direct and indirect test of sudomotor function; QPART, quantitative pilomotor axon-reflex test; SSR, sympathetic skin response.

Sympathetic skin response
Sympathetic skin response (SSR) was assessed by measuring skin conductance from medial phalanges of the index and third finger using a Powerlab polygraph (AD Instruments, Bella Vista, NSW, Australia) following sudden deep respiration to provide a comparative measure of sudomotor function as previously described [11]. The maximum increase in amplitude was calculated to provide a composite measure of preganglionic and postganglionic sympathetic sudomotor function.

Heart rate variability
In addition to the initially defined set of assessment tools, we included analysis of heart rate variability (HRV) in our study to assess a comparative predominantly parasympathetic parameter of neurocardiac function. We used a biophysiological signal analysis software package (LabChart 5; AD Instruments, Castle Hill, NSW, Australia) to compute the time-domain HRV parameters SD of NN-intervals and root mean square of successive differences between normal heartbeats following artifact removal from electrocardiogram as previously described [12]. Spectral power analysis was carried out applying a fast Fourier transformation and absolute power values were determined for low-frequency band (0.04-0.15 Hz), a parameter that is influenced by both the sympathetic and the parasympathetic system, as well as the predominantly parasympathetic high-frequency band (0.15-0.4 Hz).
Assessment of HRV was repeated under metronomic breathing at 6 cycles/min, a technique that has been shown to yield maximal parasympathetic tone.

F I G U R E 2
Modified quantitative pilomotor axon reflex test. (a) Photographs of erect hair follicle indents are taken after iontophoresis of phenylephrine to evoke axon-reflex-mediated pilomotor response with subsequent semiautomated image segmentation. (b) Image analysis was improved using an automated analysis algorithm and artifact removal procedure. Postediting images show a reduced axon-reflex area in a subject with Parkinson disease (PD; right side) and normal pilomotor response in a healthy subject (left side). The yellow inner circles in segmented images of skin pilomotor responses delineate the direct regions of iontophoretic stimulation with phenylephrine, whereas blue outlines delineate the indirect skin region of axon-reflex-mediated pilomotor erection.

Evaluation of symptoms
Severity of symptoms related to PD and autonomic symptoms were assessed using the Movement Disorder Society-sponsored revision of the Unified Parkinson's Disease Rating Scale (UPDRS) and the Scales for Outcomes in Parkinson's Disease-Autonomic questionnaire (SCOPA-AUT) [13,14].
Prior to study start, we translated the SCOPA-AUT questionnaire into the German language. As there was no official translation beforehand, we translated the questionnaire to ensure comparability between sites and increase validity. The translation process was performed as demanded by the original authors using forward and backward translation by two independent teams, with discussion of concerning differences and repetition of the process when necessary. The translated questionnaire is now available at the webpage of the Leiden University Medical Center, the Netherlands [15].
The English version of the questionnaire can be freely accessed at the Movement Disorders Society webpage [16]. Briefly, the selfcompleted questionnaire comprises 25 categorical items evaluating the following domains: gastrointestinal (seven items), urinary (six items), cardiovascular (three items), thermoregulatory (four items), pupillomotor (one item), and sexual (two items for men and two items for women). Items were rated from 0 to 3 (0 = never, 1 = sometimes, 2 = regularly, 3 = often), with higher total scores indicating more severe autonomic symptoms.

Statistical analysis
All analyses were performed using the statistical software package Stata (version 16.1; College Station, TX, USA). Outcome variables were checked for normality using descriptive (skewness, kurtosis) and analytic (Shapiro-Wilk test) criteria. Significance level was set at α = 0.05. No correction for multiple comparisons was performed due to the exploratory character of study. Outliers were removed when obviously violating physiological reasonability most probably due to instrumental bias. Missing data were not imputed. After descriptive and analytical checking for normality of continuous variables, we displayed mean or median with SD or interquartile range (IQR), respectively, where appropriate.
Pilomotor function was the primary outcome, whereas symptoms and sudomotor and neurocardiac parameters were considered secondary and other outcomes. Between-group differences were assessed with unpaired t-test or Mann-Whitney U-test accordingly.
Analyses were performed using multilevel linear mixed models for each investigated outcome variable including pilomotor and sudomotor axon-reflex function, HRV, and SSR. Main effects of group, time, and time × group interaction were computed. Time and group were declared as fixed effects. Random effects parts consist of group and patient ID with a random intercept for each. Random effect of patient ID was again a function of time, thereby adding a random slope for time at random effect level patient ID. Main effects were adjusted for covariates disease duration and age; significant correlation between age and disease duration at baseline was excluded (p = 0.4). Margins plots were used to visualize interactions.
To assess significance of differences between predictive margins of disease and control groups, contrasts of margins were plotted with control group as reference. p-Values for overall group effect, group × time interaction effect, and effect of disease duration were generated from a multilevel linear mixed model. Significance of predicted margins was derived from the mixed model. Spearman rank correlation was calculated using the complete original dataset to assess associations between pilomotor function and the UPDRS motor part as well as SCOPA-AUT score. Differences between baseline and Week 2 as well as differences of pilomotor function between the dominant and nondominant side were analyzed using Wilcoxon sign-rank test.

Standard protocol approvals, registrations, and patient consents
The study protocol has been approved by the institutional review

Charité University Medicine Berlin accepted IRB approval from
Technische Universität Dresden. The study protocol was registered on Clini calTr ials.gov (NCT03043768) and was published prior to commencement elsewhere [17]. Written and oral informed consent was obtained from each participant.

Study participants
We recruited 50 participants at all study centers. The data of 12 healthy control subjects were not accessible for technical issues in the data storage system. Thus, 38 participants (26 PD patients, 12 healthy controls) provided data for analysis. Of these, all completed baseline investigation and 2-week follow-up, whereas 30 participants completed the study investigations after 1 year and 16 after 2 years. The main reasons for low recruitment rates were the relatively broad exclusion criteria and difficulties in recruiting healthy controls due to the limited incentive for people without contact with PD. Moreover, relatives of PD patients frequently fulfilled exclusion criteria. Study groups did not differ in baseline characteristics as displayed in Table 1.

Reliability of pilomotor function assessment
Axon-reflex measures of pilomotor function showed reliability throughout all parameters with no changes from baseline to Week 2 in healthy control subjects and in PD patients as shown in Table 2.

Side-to-side comparison of pilomotor function
We observed no side-to-side differences in pilomotor function of PD patients when comparing extremities with higher motor impairment to lesser affected contralateral extremities as shown in Table S1.

Pilomotor function at baseline
Lower leg pilomotor function was impaired in PD patients compared to healthy control subjects at baseline as illustrated in Figure 3.
However, there was no difference between groups when measurements were performed on the forearms as shown in Table S2.

Longitudinal changes of pilomotor function
Pilomotor dysfunction of the lower left leg of patients with PD relative to controls increased over time, although no decline was noted in terms of absolute numbers in both groups as shown in Figure 4a.

AUTONOMIC SKIN FUNCTION IN PD
showed no baseline differences but projected relative impairment compared to healthy controls 2 years after study entry as shown in Figure S2c-f.

Association of pilomotor function and symptoms
Pilomotor dysfunction correlated with motor symptoms in patients with PD. This association was not present at baseline but became apparent as a trend after 1 year and eventually reached statistical significance after 2 years as illustrated in Figure 5. By contrast, disease duration did not correlate with pilomotor function, whereas autonomic symptoms displayed a correlation at baseline, which was not sustained at follow-ups after 1 and 2 years as shown in Table S3.
The most frequent autonomic symptoms in PD patients were gastrointestinal disturbances and urogenital function ( Table 1). Individual patient data on pilomotor function and domains of autonomic symptoms are shown in Table S4.

Sudomotor axon-reflex function
In contrast to pilomotor function assessment, sudomotor function evaluation using QDIRT showed neither any differences between healthy control subjects and patients with PD nor any changes over time as shown in Table S5.  Figure S3.

Neurocardiac function
Time-domain and spectral analysis of HRV under resting conditions and paced breathing revealed neither differences between PD patients and healthy subjects at baseline nor any changes over time in both groups as detailed in Table S6.

DISCUSS ION
A major finding of this study is that autonomic skin dysfunction in early stages of PD can be detected using QPART, most effectively when performed on the lower legs with evidence of reliability on short-term repetition. Pilomotor dysfunction correlated with motor symptoms and relative differences between PD patients and healthy subjects increased over the course of 2 years but were unexpectedly not paralleled by a decline in absolute values of pilomotor function. Pilomotor function showed no lateralization in the side-to-side comparison, consistent with a peripheral more symmetric pathogenesis of autonomic dysfunction in PD, which does not originate from asymmetric brain pathology. Although these observations might be in favor of the utility of pilomotor function studies in the evaluation of PD patients, sudomotor axon-reflex assessment did not reveal consistent differences between PD patients and healthy subjects, which might be explained by high intersubject variability.
Analysis of the sympathetic pilomotor axon-reflex response and the sudomotor SSR showed consistent impairment in PD patients contrasting with mainly parasympathetic parameters of neurocardiac function that did not display any between-group differences.
Taken together, these observations might indicate a predominant sympathetic mechanism of PD-related nerve fiber damage, which is in line with previous results from structural nerve fiber analyses showing alpha-synuclein in sympathetic nerve fibers that innervate sweat glands, blood vessels, and arrector pili muscle in the skin of TA B L E 2 Repeated pilomotor function assessment 2 weeks after baseline. Note: The indirect area of axon-reflex-mediated pilomotor erection was unchanged 2 weeks after baseline assessment, both in the group of patients with PD as well as in healthy control subjects.
PD patients [4,6,[18][19][20]. Although predictive margin analyses re- However, it is noteworthy that assessment of the SSR, a technique known for high between-subject variability and high sensitivity to changes in emotion, displayed autonomic sudomotor impairment in PD patients in our study [22]. In contrast with purely postganglionic axon-reflex tests, SSR is a composite test of preganglionic and postganglionic sympathetic function. Viewed in conjunction with our observation of unaltered sudomotor axon-reflex function, attenuated SSR in PD might indicate dysfunction of sudomotor control centers in our population of patients with early stages of PD. This interpretation is also in line with the results of a recent study of electrochemical skin conductance in patients with newly diagnosed PD where no differences from healthy subjects could be found [23]. A predominantly central mechanism of sympathetic sudomotor dysfunction might also be consistent with the results of a study in 35 patients with early-to-intermediate PD, which revealed neither peripheral sympathetic nerve fiber impairment nor changes in adrenoreceptor sensitivity but attenuated muscle sympathetic nerve activity [24].
A recent study in patients with PD and multiple system atrophyparkinsonian type (MSA-P) using quantification of local sweat response to pilocarpine iontophoresis showed distinct differences in the postganglionic component of sudomotor dysfunction between both disorders, with more pronounced impairment in MSA-P [25].
Another study was able to show that PD patients with higher disease stage tended to have impaired sudomotor axon-reflex function [24]. However, this trend did not reach statistical significance and the sudomotor axon-reflex was assessed without spatial resolution, possibly limiting comparability with our observations.
In our study, pilomotor changes were most pronounced in the lower legs. This observation is consistent with accumulating evidence that autonomic involvement in PD starts in the periphery and may support the hypothesis that alpha-synuclein pathology in PD follows a pattern of distal to proximal progression along autonomic neural pathways [6]. was performed at baseline but was not repeated over the course of the study. Thus, later manifestations of initially subclinical neuropathies cannot be ruled out. We did not correlate our observations from autonomic skin testing with structural measures of nerve fiber integrity. We did not perform QDIRT on the legs to maximize compliance during testing. Therefore, we were not able to inquire F I G U R E 5 Pilomotor dysfunction correlated with motor symptoms in Parkinson disease. Scatter plots illustrate the correlation between spatial spread of the pilomotor axon-reflex and motor symptoms assessed via Unified Parkinson's Disease Rating Scale (UPDRS) Part III. The illustrated data derived from measurements on the right leg (rho, Spearman rank correlation coefficient).
about possible subtle sudomotor changes at the site were a possible length-dependent small fiber neuropathy would become first apparent. Although our broad yet specific exclusion criteria allowed a targeted analysis of the effects of PD on autonomic skin function, potential effects of pharmacological agents acting on the autonomic nervous system require further investigation.
In conclusion, our results support the utility of small autonomic nerve fiber function assessment via QPART as a noninvasive diagnostic tool to detect PD-related autonomic neuropathy, monitor disease progression, and potentially evaluate response to diseasemodifying treatment. Furthermore, our observations indicate that adrenergic pilomotor nerve fibers display more severe affection in the course of PD than cholinergic sudomotor fibers and might be subject to adrenoreceptor hypersensitivity in early disease stages, the clinical diagnostic utility of which requires further investigation.

ACK N OWLED G M ENTS
The authors extend their sincere appreciation to Prof. Joachim Fauler for his generous support and mentorship. The authors thank Ms. Leonie Stibal for her assistance with recruitment. Open Access funding enabled and organized by Projekt DEAL.

FU N D I N G I N FO R M ATI O N
This study was funded by the Michael J. Fox Foundation for Parkinson's Research (grant ID: 10448) and Prothena Biosciences. The GWT-TUD was the sponsor of this study and contributed to study management.

CO N FLI C T O F I NTER E S T S TATEM ENT
W.Z. is a fulltime employee of Prothena Biosciences. The other authors declare no commercial or financial relationships that could be construed as a potential conflict of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.