https://orcid.org/0000-0002-1330-3321
Abstract
We aimed to describe the clinical features of patients with pure autonomic failure (PAF) preceding phenoconversion that could be useful as predictive markers for advancing α-synuclein-associated neurodegeneration of the brain.
Patients diagnosed with PAF were evaluated at eight centres (seven US-based and one European) and enrolled in a longitudinal observational cohort study (NCT01799915). Subjects underwent detailed assessments of motor, sleep, olfactory, cognitive and autonomic function and were followed prospectively to determine whether they developed parkinsonism or dementia for up to 10 years. We identified incident cases of Parkinson's disease (PD), dementia with Lewy bodies (DLB) or multiple system atrophy (MSA) and computed hazard ratios for phenoconversion as functions of clinical features.
A total of 209 participants with PAF with a median disease duration of 6 years (IQR: 3–10) were enrolled. Of those, 149 provided follow-up information at an office or telemedicine visit. After a mean follow-up duration of 3 years, 48 (33%) participants phenoconverted (42% to PD, 35% to DLB and 23% to MSA). Faster phenoconversion from study enrolment to any diagnosis was associated with urinary and sexual dysfunction [hazard ratio (HR) 5.9, 95% confidence interval (CI): 1.6–22 and HR: 3.6, 95% CI: 1.1–12] followed by subtle motor signs (HR: 2.7, 95% CI: 1.2–6), trouble swallowing (HR 2.5, 95% CI: 1.4–4.5) and changes in speech (HR:2.4, 95% CI:1.1–4.8) at enrolment. Subjects reporting deterioration of handwriting were more likely to phenoconvert to PD (HR: 2.6, 95% CI: 1.1–5.9) and those reporting difficulty handling utensils were more likely to phenoconvert to DLB (HR: 6.8, 95% CI: 1.2–38). Patients with a younger age of PAF onset (HR: 11, 95% CI: 2.6–46), preserved olfaction (HR: 8.7, 95% CI: 1.7–45), anhidrosis (HR: 1.8, 95% CI: 1–3.1, P = 0.042) and severe urinary problems (HR 1.6, 95% CI: 1–2.5, P = 0.033) were more likely to phenoconvert to MSA. The best autonomic predictor of PD was a blunted heart rate increase during the tilt-table test (HR: 6.1, 95% CI: 1.4–26).
Patients with PAF have an estimated 12% (95% CI: 9–15%) per year annual risk following study entry of phenoconverting to a manifest CNS synucleinopathy.
Introduction
Pure autonomic failure (PAF) is a prodromal ‘non-motor’ neurodegenerative synucleinopathy that often progresses to Parkinson's disease (PD), dementia with Lewy bodies (DLB) or multiple system atrophy (MSA).1-4 PAF is associated with α-synuclein deposits in neurons or glia throughout the autonomic nervous system that impair norepinephrine release from peripheral sympathetic nerve terminals innervating the vasculature and heart.5,6 The chief clinical manifestation is neurogenic orthostatic hypotension (nOH), i.e. a fall in blood pressure on standing, that increases the risk of fainting and falls. In addition, patients with PAF often have widespread signs of autonomic failure including bladder, bowel and sweat dysfunction in the absence of overt sensory deficits or autoimmune causes.1-4,7
There is now substantial evidence from retrospective and prospective studies to establish that PAF is a risk factor for PD, DLB or MSA, and research criteria for the prodromal forms of all three diseases now incorporate the manifestations of PAF as a predictive risk marker for future phenoconversion.6,7 However, it is still not known which combination of biomarkers can be used to subtype the PAF population, and accurately predict who will develop the more rapidly progressive MSA phenotype or the more slowly progressive PD/DLB.1-3,8 While the prodromal phase of the α-synucleinopathies from the perspective of REM sleep behaviour disorder (RBD) and anosmia is well described,4 there is still a lack of long-term well powered prospective natural history data in PAF. To overcome this, the Autonomic Disorders Consortium was founded in 2008, as a multi-centre collaborative effort with a core project to recruit and follow patients with PAF to understand the prognostic implications of nOH.
Our initial report of the PAF cohort showed that patients with PAF had a high risk of phenoconverting to PD, DLB or MSA, but the small original sample size of 74 patients lacked the power to predict those most at risk of fast phenoconversion, as a substantial number of patients retained the PAF phenotype for several decades.1 This long prodromal phase and the absence of predictive biomarkers hampers our ability to assemble an enriched cohort of PAF patients for prodromal clinical trials and select precise end points to measure motor and/or cognitive decline.
Now with up to 10 years of prospective data, we present an update to the 2017 natural history study of PAF,1 with a substantially larger cohort followed for an additional 6 years. We combine multiple clinical markers to predict clinical trajectories in patients who present with PAF and the risk of rapid neurological decline. The results should help support the identification of an enriched homogeneous prodromal population to test potential neuroprotective strategies for the synucleinopathies aimed at biologically targeting misfolded aggregated α-synuclein deposits.8
Materials and methods
Participants and protocol
Between September 2011 and March 2022, patients diagnosed with PAF and followed within the Autonomic Disorders Consortium's Natural History Study of the Synucleinopathies (NCT01799915) were evaluated at seven academic autonomic centres in the USA (New York University Langone Medical Center, New York, NY; Vanderbilt University, Nashville, TN; Mayo Clinic Rochester, MN; NIH Intramural Research Program, Bethesda, MD; Beth Israel Deaconess Medical Center, Boston, MA; Stanford University, Palo Alto, CA; University of Texas Southwestern, Dallas, TX) and one in Europe (Medical University of Innsbruck, Innsbruck, Austria). At enrolment, participants underwent comprehensive neurological examinations to exclude substantial motor or sensory deficits, dementia or other causes of autonomic neuropathy.7 Autonomic function testing was performed with continuous monitoring of blood pressure and electrocardiographic R-R intervals. The diagnosis of PAF was based on current criteria7 including (i) nOH defined as a drop in systolic/diastolic blood pressure within 3 min upright of more than 20/10 mmHg; and (ii) absence of the phase IV blood pressure overshoot after the release of the Valsalva strain; or (iii) in patients who were unable to adequately perform the Valsalva manoeuvre, a heart rate-to-systolic blood pressure ratio (ΔHR/ΔSBP) < 0.5 bpm/mmHg in response to orthostatic stress.9,10 The main exclusion criteria were clinically significant peripheral sensory or motor neuropathy, diagnosis of another disease known to cause peripheral autonomic neuropathy (e.g. diabetes mellitus, amyloidosis, toxic neuropathy, autoimmune ganglionopathy or other immune-mediated/paraneoplastic syndromes), fulfilment of diagnostic criteria for PD, MSA or DLB at entry, or other non-neurogenic cause for OH including severe anaemia, heart failure, intravascular volume depletion or OH secondary to medication side effects. All 74 PAF patients in the original study were included in this updated analysis. Local institutional review boards provided approval for all study procedures at each site and informed consent was obtained from all participants. During the COVID-19 pandemic, the protocol was amended to include data collected through remote telemedicine visits in participants who were unable to be evaluated in person.
Clinical evaluation
Participants underwent standardized assessments at the time of enrolment and each annual follow-up visit. At each visit, we performed a complete neurological examination documenting cognitive, parkinsonian, cerebellar and pyramidal manifestations. Olfactory function was quantified using the University of Pennsylvania Identification Test (UPSIT).11 Cognitive function was screened using the Montreal Cognitive Assessment (MoCA), corrected for education <12 years (+1 point), considering normal scores of 26 and above, mild cognitive impairment (MCI) as a score of 18–25, moderate cognitive impairment as a score of 10–17 and severe cognitive impairment as a score of 10 and below.12 Dream-enactment behaviour was assessed using the validated RBD-single question and, where possible, confirmation by polysomnography.13 The severity of motor dysfunction was scored using the Unified Multiple System Atrophy Rating Scale (UMSARS)14 using self-reported scores of symptom-related disabilities (UMSARS-I), clinical assessment of motor deficits (UMSARS-II) and a global disability score (UMSARS-IV). Medication use and response to treatment were collected annually. Autonomic function testing was standardized across sites and included a battery of established tests. Before testing, subjects were instructed to withhold anti-hypotension medications for 48 h or five half-lives, whichever was longest, arrive fasted and avoid caffeine and alcohol.
Cardiac parasympathetic (vagal) function was evaluated in the time and frequency domain using heart rate variability measures. Haemodynamic changes, induced by a Valsalva manoeuvre and head-up tilt or active standing, were obtained. As an indicator of baroreflex-mediated sympathetic activation, supine and standing plasma catecholamine were sampled through an indwelling catheter and assayed by high-performance liquid chromatography at each local site.1 The burden of systemic autonomic symptoms was assessed using the COMPASS-31 scale.15
Follow-up and end points
To enhance subject retention, all efforts were made to re-contact participants missing their annual follow-up visits via phone, e-mail and postal mail. Participants who were unable to be reached after multiple attempts were considered lost to follow-up. Subjects were encouraged to return for in-person follow-up visits with full autonomic testing, and when not possible, telemedicine visits were performed with measurements of blood pressure and heart obtained during a guided active standing test at home.
Phenoconversion was defined as a new diagnosis of PD, DLB or MSA (incident case) in participants who were enrolled as PAF, and subsequently met current clinical diagnostic criteria for PD, DLB or MSA at any follow-up visit.16-18 Subtle motor signs in the PAF phase (before phenoconversion) were noted if the subject had mild findings on motor examination, such as mild generalized slowness/bradykinesia, decreased blinking frequency or reduced facial expressions (minimal hypomimia), reduced arm swing when walking, or mild slowing/reduction in amplitude in rapid alternating movements, not sufficient to establish a diagnosis of PD, DLB or MSA and in the opinion of the investigator, were not confounded by concomitant medical conditions such as arthritis, orthopaedic issues or cerebrovascular disease.19 Retention of the PAF phenotype was defined as participants who did not develop significant resting tremor, overt bradykinesia, ataxia or dementia at the time of their last visit.
Age at PAF onset was defined as the age at which the subject first recalled noticing symptoms of nOH, such as dizziness, light-headedness, coat-hanger pain or syncope on standing. Supine hypertension was defined as a blood pressure >140/90 mmHg in the supine position after at least 5 min of rest.20
Data management and statistics
Data were collected on standardized electronic source documents available to all sites and managed by the New York University Data Coordinating Center, using the REDCap platform.21 Normality was assessed using the Shapiro-Wilk test. Position and dispersion measures are expressed as mean ± standard deviation (SD) for parametric variables and as median and interquartile range (IQR) for non-parametric variables. Time-to-event analyses were conducted using the time origins of (i) study enrolment (duration of follow-up); (ii) onset of PAF (duration of PAF); and (iii) birth (age at phenoconversion). Events of interest were (i) phenoconversion to any diagnosis; (ii) phenoconversion to PD; (iii) phenoconversion to MSA; (iv) phenoconversion to DLB; and (v) phenoconversion to either PD or DLB. For a given event, patients were treated as censored at the minimum of the end of their follow-up and their diagnosis with an alternative event (e.g. a patient diagnosed with PD is censored in the time to MSA analysis at the time of PD diagnosis). The use of onset of PAF or birth as the time origins requires adjustment for left truncation22 (i.e. delayed entry), to account for the requirement that patients have not yet phenoconverted at enrolment to the study. Without this adjustment, the times to phenoconversion would be biased relative to the population of all patients who experience the onset of PAF, with longer times to phenoconversion. Kaplan-Meier curves were used to estimate time-to-event distributions, with adjustment for left truncation for estimates from onset of PAF and from birth. Cox regression models for time from study enrolment to diagnosis were fit, with adjustment for the duration of PAF at enrolment. These models are well suited to our data as they do not lose information through a truncation adjustment and appropriately deal with time-varying covariates that are measured only at study entry.23 We did not adjust for multiple testing and provide 95% confidence intervals (CI) for estimation. Data were analysed using SPSS v23.0 (IBM, Armonk, New York) and Rstudio 23.03.0.
Results
Clinical characteristics at enrolment
At study entry, the cohort included 209 participants with PAF, with 133 (64%) males and 76 (36%) females, and a median age of 70 years (IQR 65–76, range 41–88). The cohort included all 74 PAF subjects that were described in our previous study,1 with the addition of 135 additional subjects recruited since 2015 (Supplementary Table 1).
Almost all participants (92%) were fully or almost fully independent, with 4% being moderately and 4% very dependent, as measured by the UMSARS-IV.
The median age of onset of PAF was 65 years (IQR: 58–70, range 42–88), with a median disease duration of 6 years at enrolment (IQR 3–10, range 1–17).
Orthostatic blood pressure and heart rate were obtained by tilt table test in 156 (75%) subjects and by active standing in the remaining 53 subjects (25%). The median supine blood pressure was 155/83 mmHg (IQR: 137/73–177/95 mmHg), with a heart rate of 64 bpm (IQR: 60–71 bpm). After 3 min upright, blood pressure was reduced on average to 103/62 mmHg (IQR: 84/51–128/74 mmHg), which was accompanied by a blunted increase in heart rate to 74 bpm (IQR: 67–84 bpm). The median orthostatic fall in blood pressure after 3 min upright was 49/18 mmHg (IQR 31/9–66/31 mmHg). Despite the fall in blood pressure, heart rate increased by only 9 bpm (IQR 3–15 bpm). Blood pressure continued to fall throughout the 10 min of orthostatic stress, and the lowest blood pressure captured was on average 85/53 mmHg (IQR: 72/46–100/65 mmHg). All participants met the diagnostic criteria for nOH with a > 20/10 mmHg fall in blood pressure within 3 min of tilt or active standing. Blood pressure overshoot after release of the Valsalva strain was absent in all 197 subjects that could perform the manoeuvre, and all 12 patients (6%) that could not perform a Valsalva manoeuvre, all showed a blunted ΔHR/ΔSBP ratio (0.15 bpm/mmHg, IQR: 0.04–0.3 bpm/mmHg).
The median ΔHR/ΔSBP ratio was 0.14 bpm/mmHg (IQR: 0.05–0.28 bpm/mmHg), consistent with a neurogenic cause for the OH.9,10 Supine plasma norepinephrine levels showed a wide range (IQR: 85–270 pg/ml, range 5–1321 pg/ml), but were on average on the lower end than expected for elderly adults (135 pg/ml). Plasma norepinephrine increased by only 53% after 10 min upright (22–110%), which is consistent with impaired baroreflex activation of sympathetic noradrenergic neurons during hypotension.24
Motor and non-motor features at enrolment
Despite none of the participants meeting the clinical diagnostic criteria for PD, MSA or DLB on study entry, 143 (68%) had evidence of subtle motor signs on careful neurological examination. Sixty-five (31%) had a mildly abnormal posture or gait, 51 (24%) had mild high-frequency action tremor, 47 (22%) had slow speech or reduced facial expression, 44 (21%) had mild bradykinesia, 44 (21%) had mild rigidity, 42 (20%) had mild incoordination and 24 (14%) had self-reported ‘movement difficulty’. None had a resting ‘pill-rolling’ tremor. Polysomnography tests were performed in 29/209 subjects (14%). In total, 116 participants (56%) reported suspected dream enactment behaviour on a questionnaire and gave histories suggestive of clinical RBD. RBD was confirmed by polysomnography testing in 27 subjects that endorsed RBD by questionnaire and excluded in the two subjects that denied RBD activity on the questionnaire but underwent polysomnography.
The median MoCA score at study entry was 27 points (IQR: 25–30), with 48 participants (26%) meeting the cut-off for MCI and seven participants (4%) having moderate cognitive impairment. No participants had dementia.25 Most participants had hyposmia (47%) or anosmia (39%) and, on average, the mean UPSIT score was within the hyposmic range (22 points, IQR: 15–29).
Longitudinal follow-up
Supplementary Fig. 1 shows the participant flow, phenoconversions, deaths and losses to follow-up each year. One hundred and forty-six participants (70%) completed the baseline (enrolment) visit and a follow-up visit 12 months after. Table 1 describes the clinical and autonomic characteristics at enrolment of participants who phenoconverted and those that did not at their last available follow-up (i.e. remained as PAF).
Characteristics at enrolment in phenoconverters and those that retained the PAF diagnosis at their final visit
| Remained PAF n = 161 | All phenoconverters n = 48 | PD (n = 20) | DLB (n = 17) | MSA (n = 11) | |
|---|---|---|---|---|---|
| Sex: male/female (n) | 99/62 | 34/14 | 16/4 | 12/5 | 6/5 |
| Age at onset (years) | 64 (57–69) | 66 (59–71) | 63 (59–70) | 69 (66–76) | 57 (54–66) |
| Duration of PAF (years) | 6 (4–10) | 4 (2–7) | 8 (5–13) | 7 (6–11) | 5 (3–7) |
| Phenoconversion age (years) | N/A | 74 (70–78) | 75 (72–78) | 78 (74–80) | 65 (56–69) |
| Clinical RBD (n, %) | 85 (53%) | 31 (65%) | 13 (65%) | 10 (59%) | 8 (73%) |
| MoCA | 27 (24–29) | 30 (28–30) | 30 (29–30) | 29 (26–30) | 29 (28–30) |
| UPSIT (points) | 22 (16–29) | 20 (14–28) | 17 (15–21) | 18 (11–23) | 33 (30–35) |
| Normosmia | 20/131 (15%) | 5/39 (13%) | 1/16 (6%) | 0/14 (0%) | 4/9 (44%) |
| Anosmia | 50/131 (38%) | 17/39 (44%) | 10/16 (63%) | 7/14 (50%) | 0/9 (0%) |
| Subtle motor signs | 109 (68%) | 34 (71%) | 16 (80%) | 10 (59%) | 8 (73%) |
| UMSARS-I | 7 (5–10) | 9 (6–12) | 8 (6–12) | 9 (6–13) | 9 (6–12) |
| UMSARS-II | 1 (0–3) | 2 (0–5) | 1 (0–11) | 3 (0–5) | 3 (0–5) |
| Autonomic markers | |||||
| E:I ratio | 1.06 (1.03–1.10) | 1.08 (1.05–1.15) | 1.07 (1.06–1.11) | 1.10 (1.04–1.23) | 1.07 (1.04–1.10) |
| Supine BP (mmHg) | 156/83 (137/73–177/95) | 153/84 (137/74–178/95) | 147/83 (137/74–172/94) | 153/79 (130/73–178/86) | 168/89 (138/84–182/100) |
| Supine HTN, (n, %) | 118 (73%) | 33 (69%) | 15 (75%) | 11 (65%) | 7 (64%) |
| Supine heart rate (bpm) | 64 (60–71) | 65 (60–71) | 63 (60–67) | 63 (59–72) | 70 (65–73) |
| 3-min standing BP (mmHg) | 105/62 (86/52–129/75) | 99/55 (78/47–119/71) | 91/53 (78/48–107/64) | 97/64 (86/48–127/77) | 118/60 (94/55–131/83) |
| 3-min standing HR (bpm) | 74 (67–84) | 74 (66–87) | 71 (65–77) | 76 (67–86) | 82 (72–88) |
| ΔHR 3-min stand (bpm) | 9 (3–15) | 9 (4–14) | 5 (2–10) | 9 (7–17) | 13 (10–15) |
| ΔHR/ ΔSBP (bpm/mmHg) | 0.14 (0.06–0.28) | 0.15 (0.08–0.31) | 0.08 (0.04–0.14) | 0.16 (0.14–0.36) | 0.21 (0.06–0.38) |
| Symptomatic nOH | 81 (68%) | 22 (63%) | 10 (71%) | 9 (64%) | 3 (43%) |
| NE supine (pg/ml) | 127 (76–277) | 155 (102–222) | 159 (88–226) | 136 (98–235) | 141 (133–193) |
| Δ NE (%) | 50% (21–110%) | 60% (23–92%) | 53% (3586%) | 82% (18–174%) | 54% (21–70%) |
| COMPASS-31 total score | 37.1 (26.2–47.4) | 33.7 (0–42.5) | 33.8 (28.4–39.7) | 17.2 (0–38.4) | 36.5 (0–54.7) |
| COMPASS-31GI score | 7.1 (2.7–9.8) | 0 (0–8) | 5.4 (0–7.1) | 0 (0–4.5) | 1.3 (0–12.5) |
| Laxative use | 42 (32%) | 10 (29%) | 4 (29%) | 3 (25%) | 3 (33%) |
| COMPASS-31Urinary score | 1.1 (0–2.2) | 3.3 (2.2–5.6) | 2.2 (1.1–3.3) | 4.4 (3.3–5.6) | 5 (2.2–8.3) |
| Catheterization | 8 (6%) | 3 (8%) | 0 (0%) | 1 (8%) | 2 (20%) |
| COMPASS-31 SM score | 4.3 (2.1–6.4) | 4.3 (2.1–6.4) | 2.1 (0–4.3) | 4.3 (4.3–4.3) | 5.4 (1.1–9.6) |
| Sexual dysfunction | 88 (71%) | 26 (76%) | 11 (79%) | 6 (60%) | 9 (90%) |
| Remained PAF n = 161 | All phenoconverters n = 48 | PD (n = 20) | DLB (n = 17) | MSA (n = 11) | |
|---|---|---|---|---|---|
| Sex: male/female (n) | 99/62 | 34/14 | 16/4 | 12/5 | 6/5 |
| Age at onset (years) | 64 (57–69) | 66 (59–71) | 63 (59–70) | 69 (66–76) | 57 (54–66) |
| Duration of PAF (years) | 6 (4–10) | 4 (2–7) | 8 (5–13) | 7 (6–11) | 5 (3–7) |
| Phenoconversion age (years) | N/A | 74 (70–78) | 75 (72–78) | 78 (74–80) | 65 (56–69) |
| Clinical RBD (n, %) | 85 (53%) | 31 (65%) | 13 (65%) | 10 (59%) | 8 (73%) |
| MoCA | 27 (24–29) | 30 (28–30) | 30 (29–30) | 29 (26–30) | 29 (28–30) |
| UPSIT (points) | 22 (16–29) | 20 (14–28) | 17 (15–21) | 18 (11–23) | 33 (30–35) |
| Normosmia | 20/131 (15%) | 5/39 (13%) | 1/16 (6%) | 0/14 (0%) | 4/9 (44%) |
| Anosmia | 50/131 (38%) | 17/39 (44%) | 10/16 (63%) | 7/14 (50%) | 0/9 (0%) |
| Subtle motor signs | 109 (68%) | 34 (71%) | 16 (80%) | 10 (59%) | 8 (73%) |
| UMSARS-I | 7 (5–10) | 9 (6–12) | 8 (6–12) | 9 (6–13) | 9 (6–12) |
| UMSARS-II | 1 (0–3) | 2 (0–5) | 1 (0–11) | 3 (0–5) | 3 (0–5) |
| Autonomic markers | |||||
| E:I ratio | 1.06 (1.03–1.10) | 1.08 (1.05–1.15) | 1.07 (1.06–1.11) | 1.10 (1.04–1.23) | 1.07 (1.04–1.10) |
| Supine BP (mmHg) | 156/83 (137/73–177/95) | 153/84 (137/74–178/95) | 147/83 (137/74–172/94) | 153/79 (130/73–178/86) | 168/89 (138/84–182/100) |
| Supine HTN, (n, %) | 118 (73%) | 33 (69%) | 15 (75%) | 11 (65%) | 7 (64%) |
| Supine heart rate (bpm) | 64 (60–71) | 65 (60–71) | 63 (60–67) | 63 (59–72) | 70 (65–73) |
| 3-min standing BP (mmHg) | 105/62 (86/52–129/75) | 99/55 (78/47–119/71) | 91/53 (78/48–107/64) | 97/64 (86/48–127/77) | 118/60 (94/55–131/83) |
| 3-min standing HR (bpm) | 74 (67–84) | 74 (66–87) | 71 (65–77) | 76 (67–86) | 82 (72–88) |
| ΔHR 3-min stand (bpm) | 9 (3–15) | 9 (4–14) | 5 (2–10) | 9 (7–17) | 13 (10–15) |
| ΔHR/ ΔSBP (bpm/mmHg) | 0.14 (0.06–0.28) | 0.15 (0.08–0.31) | 0.08 (0.04–0.14) | 0.16 (0.14–0.36) | 0.21 (0.06–0.38) |
| Symptomatic nOH | 81 (68%) | 22 (63%) | 10 (71%) | 9 (64%) | 3 (43%) |
| NE supine (pg/ml) | 127 (76–277) | 155 (102–222) | 159 (88–226) | 136 (98–235) | 141 (133–193) |
| Δ NE (%) | 50% (21–110%) | 60% (23–92%) | 53% (3586%) | 82% (18–174%) | 54% (21–70%) |
| COMPASS-31 total score | 37.1 (26.2–47.4) | 33.7 (0–42.5) | 33.8 (28.4–39.7) | 17.2 (0–38.4) | 36.5 (0–54.7) |
| COMPASS-31GI score | 7.1 (2.7–9.8) | 0 (0–8) | 5.4 (0–7.1) | 0 (0–4.5) | 1.3 (0–12.5) |
| Laxative use | 42 (32%) | 10 (29%) | 4 (29%) | 3 (25%) | 3 (33%) |
| COMPASS-31Urinary score | 1.1 (0–2.2) | 3.3 (2.2–5.6) | 2.2 (1.1–3.3) | 4.4 (3.3–5.6) | 5 (2.2–8.3) |
| Catheterization | 8 (6%) | 3 (8%) | 0 (0%) | 1 (8%) | 2 (20%) |
| COMPASS-31 SM score | 4.3 (2.1–6.4) | 4.3 (2.1–6.4) | 2.1 (0–4.3) | 4.3 (4.3–4.3) | 5.4 (1.1–9.6) |
| Sexual dysfunction | 88 (71%) | 26 (76%) | 11 (79%) | 6 (60%) | 9 (90%) |
BP = blood pressure; bpm = beats per minute; COMPASS = Composite Autonomic Symptom Score; DLB = dementia with Lewy bodies; E:I ratio = expiratory to inspiratory ratio; GI = gastrointestinal; HR = heart rate; HTN = hypertension; MoCA = Montreal Cognitive Assessment; MSA = multiple system atrophy; NE = norepinephrine; nOH = neurogenic orthostatic hypotension; PAF = pure autonomic failure; PD = Parkinson’s disease; RBD = REM sleep behaviour disorder; SM = secretomotor; SBP = systolic blood pressure; UMSARS = Unified Multiple System Atrophy Rating Scale; UPSIT = University of Pennsylvania Smell Identification Test.
Characteristics at enrolment in phenoconverters and those that retained the PAF diagnosis at their final visit
| Remained PAF n = 161 | All phenoconverters n = 48 | PD (n = 20) | DLB (n = 17) | MSA (n = 11) | |
|---|---|---|---|---|---|
| Sex: male/female (n) | 99/62 | 34/14 | 16/4 | 12/5 | 6/5 |
| Age at onset (years) | 64 (57–69) | 66 (59–71) | 63 (59–70) | 69 (66–76) | 57 (54–66) |
| Duration of PAF (years) | 6 (4–10) | 4 (2–7) | 8 (5–13) | 7 (6–11) | 5 (3–7) |
| Phenoconversion age (years) | N/A | 74 (70–78) | 75 (72–78) | 78 (74–80) | 65 (56–69) |
| Clinical RBD (n, %) | 85 (53%) | 31 (65%) | 13 (65%) | 10 (59%) | 8 (73%) |
| MoCA | 27 (24–29) | 30 (28–30) | 30 (29–30) | 29 (26–30) | 29 (28–30) |
| UPSIT (points) | 22 (16–29) | 20 (14–28) | 17 (15–21) | 18 (11–23) | 33 (30–35) |
| Normosmia | 20/131 (15%) | 5/39 (13%) | 1/16 (6%) | 0/14 (0%) | 4/9 (44%) |
| Anosmia | 50/131 (38%) | 17/39 (44%) | 10/16 (63%) | 7/14 (50%) | 0/9 (0%) |
| Subtle motor signs | 109 (68%) | 34 (71%) | 16 (80%) | 10 (59%) | 8 (73%) |
| UMSARS-I | 7 (5–10) | 9 (6–12) | 8 (6–12) | 9 (6–13) | 9 (6–12) |
| UMSARS-II | 1 (0–3) | 2 (0–5) | 1 (0–11) | 3 (0–5) | 3 (0–5) |
| Autonomic markers | |||||
| E:I ratio | 1.06 (1.03–1.10) | 1.08 (1.05–1.15) | 1.07 (1.06–1.11) | 1.10 (1.04–1.23) | 1.07 (1.04–1.10) |
| Supine BP (mmHg) | 156/83 (137/73–177/95) | 153/84 (137/74–178/95) | 147/83 (137/74–172/94) | 153/79 (130/73–178/86) | 168/89 (138/84–182/100) |
| Supine HTN, (n, %) | 118 (73%) | 33 (69%) | 15 (75%) | 11 (65%) | 7 (64%) |
| Supine heart rate (bpm) | 64 (60–71) | 65 (60–71) | 63 (60–67) | 63 (59–72) | 70 (65–73) |
| 3-min standing BP (mmHg) | 105/62 (86/52–129/75) | 99/55 (78/47–119/71) | 91/53 (78/48–107/64) | 97/64 (86/48–127/77) | 118/60 (94/55–131/83) |
| 3-min standing HR (bpm) | 74 (67–84) | 74 (66–87) | 71 (65–77) | 76 (67–86) | 82 (72–88) |
| ΔHR 3-min stand (bpm) | 9 (3–15) | 9 (4–14) | 5 (2–10) | 9 (7–17) | 13 (10–15) |
| ΔHR/ ΔSBP (bpm/mmHg) | 0.14 (0.06–0.28) | 0.15 (0.08–0.31) | 0.08 (0.04–0.14) | 0.16 (0.14–0.36) | 0.21 (0.06–0.38) |
| Symptomatic nOH | 81 (68%) | 22 (63%) | 10 (71%) | 9 (64%) | 3 (43%) |
| NE supine (pg/ml) | 127 (76–277) | 155 (102–222) | 159 (88–226) | 136 (98–235) | 141 (133–193) |
| Δ NE (%) | 50% (21–110%) | 60% (23–92%) | 53% (3586%) | 82% (18–174%) | 54% (21–70%) |
| COMPASS-31 total score | 37.1 (26.2–47.4) | 33.7 (0–42.5) | 33.8 (28.4–39.7) | 17.2 (0–38.4) | 36.5 (0–54.7) |
| COMPASS-31GI score | 7.1 (2.7–9.8) | 0 (0–8) | 5.4 (0–7.1) | 0 (0–4.5) | 1.3 (0–12.5) |
| Laxative use | 42 (32%) | 10 (29%) | 4 (29%) | 3 (25%) | 3 (33%) |
| COMPASS-31Urinary score | 1.1 (0–2.2) | 3.3 (2.2–5.6) | 2.2 (1.1–3.3) | 4.4 (3.3–5.6) | 5 (2.2–8.3) |
| Catheterization | 8 (6%) | 3 (8%) | 0 (0%) | 1 (8%) | 2 (20%) |
| COMPASS-31 SM score | 4.3 (2.1–6.4) | 4.3 (2.1–6.4) | 2.1 (0–4.3) | 4.3 (4.3–4.3) | 5.4 (1.1–9.6) |
| Sexual dysfunction | 88 (71%) | 26 (76%) | 11 (79%) | 6 (60%) | 9 (90%) |
| Remained PAF n = 161 | All phenoconverters n = 48 | PD (n = 20) | DLB (n = 17) | MSA (n = 11) | |
|---|---|---|---|---|---|
| Sex: male/female (n) | 99/62 | 34/14 | 16/4 | 12/5 | 6/5 |
| Age at onset (years) | 64 (57–69) | 66 (59–71) | 63 (59–70) | 69 (66–76) | 57 (54–66) |
| Duration of PAF (years) | 6 (4–10) | 4 (2–7) | 8 (5–13) | 7 (6–11) | 5 (3–7) |
| Phenoconversion age (years) | N/A | 74 (70–78) | 75 (72–78) | 78 (74–80) | 65 (56–69) |
| Clinical RBD (n, %) | 85 (53%) | 31 (65%) | 13 (65%) | 10 (59%) | 8 (73%) |
| MoCA | 27 (24–29) | 30 (28–30) | 30 (29–30) | 29 (26–30) | 29 (28–30) |
| UPSIT (points) | 22 (16–29) | 20 (14–28) | 17 (15–21) | 18 (11–23) | 33 (30–35) |
| Normosmia | 20/131 (15%) | 5/39 (13%) | 1/16 (6%) | 0/14 (0%) | 4/9 (44%) |
| Anosmia | 50/131 (38%) | 17/39 (44%) | 10/16 (63%) | 7/14 (50%) | 0/9 (0%) |
| Subtle motor signs | 109 (68%) | 34 (71%) | 16 (80%) | 10 (59%) | 8 (73%) |
| UMSARS-I | 7 (5–10) | 9 (6–12) | 8 (6–12) | 9 (6–13) | 9 (6–12) |
| UMSARS-II | 1 (0–3) | 2 (0–5) | 1 (0–11) | 3 (0–5) | 3 (0–5) |
| Autonomic markers | |||||
| E:I ratio | 1.06 (1.03–1.10) | 1.08 (1.05–1.15) | 1.07 (1.06–1.11) | 1.10 (1.04–1.23) | 1.07 (1.04–1.10) |
| Supine BP (mmHg) | 156/83 (137/73–177/95) | 153/84 (137/74–178/95) | 147/83 (137/74–172/94) | 153/79 (130/73–178/86) | 168/89 (138/84–182/100) |
| Supine HTN, (n, %) | 118 (73%) | 33 (69%) | 15 (75%) | 11 (65%) | 7 (64%) |
| Supine heart rate (bpm) | 64 (60–71) | 65 (60–71) | 63 (60–67) | 63 (59–72) | 70 (65–73) |
| 3-min standing BP (mmHg) | 105/62 (86/52–129/75) | 99/55 (78/47–119/71) | 91/53 (78/48–107/64) | 97/64 (86/48–127/77) | 118/60 (94/55–131/83) |
| 3-min standing HR (bpm) | 74 (67–84) | 74 (66–87) | 71 (65–77) | 76 (67–86) | 82 (72–88) |
| ΔHR 3-min stand (bpm) | 9 (3–15) | 9 (4–14) | 5 (2–10) | 9 (7–17) | 13 (10–15) |
| ΔHR/ ΔSBP (bpm/mmHg) | 0.14 (0.06–0.28) | 0.15 (0.08–0.31) | 0.08 (0.04–0.14) | 0.16 (0.14–0.36) | 0.21 (0.06–0.38) |
| Symptomatic nOH | 81 (68%) | 22 (63%) | 10 (71%) | 9 (64%) | 3 (43%) |
| NE supine (pg/ml) | 127 (76–277) | 155 (102–222) | 159 (88–226) | 136 (98–235) | 141 (133–193) |
| Δ NE (%) | 50% (21–110%) | 60% (23–92%) | 53% (3586%) | 82% (18–174%) | 54% (21–70%) |
| COMPASS-31 total score | 37.1 (26.2–47.4) | 33.7 (0–42.5) | 33.8 (28.4–39.7) | 17.2 (0–38.4) | 36.5 (0–54.7) |
| COMPASS-31GI score | 7.1 (2.7–9.8) | 0 (0–8) | 5.4 (0–7.1) | 0 (0–4.5) | 1.3 (0–12.5) |
| Laxative use | 42 (32%) | 10 (29%) | 4 (29%) | 3 (25%) | 3 (33%) |
| COMPASS-31Urinary score | 1.1 (0–2.2) | 3.3 (2.2–5.6) | 2.2 (1.1–3.3) | 4.4 (3.3–5.6) | 5 (2.2–8.3) |
| Catheterization | 8 (6%) | 3 (8%) | 0 (0%) | 1 (8%) | 2 (20%) |
| COMPASS-31 SM score | 4.3 (2.1–6.4) | 4.3 (2.1–6.4) | 2.1 (0–4.3) | 4.3 (4.3–4.3) | 5.4 (1.1–9.6) |
| Sexual dysfunction | 88 (71%) | 26 (76%) | 11 (79%) | 6 (60%) | 9 (90%) |
BP = blood pressure; bpm = beats per minute; COMPASS = Composite Autonomic Symptom Score; DLB = dementia with Lewy bodies; E:I ratio = expiratory to inspiratory ratio; GI = gastrointestinal; HR = heart rate; HTN = hypertension; MoCA = Montreal Cognitive Assessment; MSA = multiple system atrophy; NE = norepinephrine; nOH = neurogenic orthostatic hypotension; PAF = pure autonomic failure; PD = Parkinson’s disease; RBD = REM sleep behaviour disorder; SM = secretomotor; SBP = systolic blood pressure; UMSARS = Unified Multiple System Atrophy Rating Scale; UPSIT = University of Pennsylvania Smell Identification Test.
During follow-up, 48 of the 146 participants (33%) phenoconverted, receiving a new diagnosis of PD (n = 20), DLB (n = 17) or MSA [n = 11, seven MSA-P (parkinsonian type), four MSA-C (cerebellar type)] with phenoconversion occurring a median of 6 years after enrolment (95% CI: 5–NA). Figure 1 shows the unadjusted Kaplan-Meier curves that depict phenoconversion on the follow-up time scale, with the vast majority of patients phenoconverting to a Lewy body disorder (PD or DLB). Fifteen participants died during the follow-up period, with the most commonly known cause of death being cardiac arrest.
Phenoconversion rates according to diagnosis end point. (A) Kaplan-Meier survival graph showing phenoconversion from time from study entry (years). (B) Kaplan-Meier survival graph showing phenoconversions since the onset of pure autonomic failure (PAF) (years), with adjustments for left truncation (delayed entry). The number of subjects at risk is not always decreasing due to the definition of risk sets in which subjects are not ‘at risk’ until the onset of their symptoms of PAF. The numbers at risk are the same for each survival curve, as these include the numbers of events and numbers remaining at risk at each time point. The breakdowns of events versus numbers remaining at risk are different for the three end points. DLB = dementia with Lewy bodies; MSA = multiple system atrophy; PD = Parkinson’s disease
A younger age at PAF onset (<57 years) was associated with a higher risk of phenoconversion to MSA [hazard ratio (HR): 11, 95% CI: 2.6–46]. In our data, the unadjusted median age at phenoconversion to any diagnosis was 74 (IQR: 69–78, range 49–83), 75 years for PD (IQR 72–78, range 63–82), 78 for DLB (IQR 74–80, range 64–83) and 65 for MSA (IQR 59–69, range 49–75). These should be adjusted for left truncation by age at study enrolment, however, these estimates are not reliable due to the very small resulting risk sets. Specifically, after adjusting for left truncation by age at study entry, the median age at phenoconversion was younger for incident MSA cases (49 years, 95% CI: 49–NA) and older for both incident PD (78 years 95% CI: 75–NA) and DLB cases (83 years 95% CI: 79–NA). The median time from onset of PAF to phenoconversion to any diagnosis was 7 years (95% CI: 5–12) and the median time from PAF onset to phenoconversion to either PD or DLB was 11 years (95% CI: 7–NA); median times to other diagnoses were not estimable. The median times from study enrolment to phenoconversion was 6 years (95% CI: 5–NA); median times to individual diagnoses were not estimable (Fig. 1). The annual risk of phenoconversion from the time of study entry as PAF was 12% (95% CI: 9–15%) per year. Supplementary Table 2 summarizes the hazard ratios for the risk of phenoconversion following study entry according to the final diagnosis.
Clinical markers of phenoconversion
Motor features
As shown in Fig. 2A, the risk of phenoconversion to any diagnosis was associated with the presence of subtle motor signs including mild bradykinesia, minimal hypomimia and asymmetry as well as a higher motor rating scale score on examination at enrolment (UMSARS-2 > 3 HR: 2.7, 95% CI: 1.2–6). Specifically, early deterioration in handwriting was predictive of a higher relative risk of future PD (HR: 2.6 95% CI: 1.1–5.9, P = 0.023), whereas the best early subtle motor sign to predict a DLB outcome was difficulty handling utensils (HR: 6.8 95% CI: 1.2–38, Supplementary Fig. 2). Subtle changes in speech, such as dysarthria/lower volume/slower speech (UMSARS-1.1, HR 2.4, 95% CI: 1.1–4.8) and mild dysphagia (UMSARS-1.2, HR: 2.5, 95%CI: 1.4–4.5) were associated with a higher risk of phenoconversion to any diagnosis (Fig. 2A).
Risk factors for phenoconversion to any diagnosis. Forest plot showing hazard ratios for (A) phenoconversion to all diagnoses (PD, DLB and MSA), (B) for retaining the pure autonomic failure (PAF) phenotype at last visit, (C) for phenoconversion to MSA, and (D) for phenoconversion to Lewy body disorders (PD and DLB). The shaded areas represent a higher risk. Speech was derived from UMSARS-1.1, dysphagia was derived from UMSARS-1.2, handwriting was derived from UMSARS-1.3, handling utensils was derived from UMSARS-1.4, walking was derived from UMSARS-1.7, falls was derived from UMSARS-1.8, sexual dysfunction was derived from UMSARS-1.11, ‘bladder symptoms’ was derived from COMPASS-31 urinary subscore, ‘secretomotor function’ was derived from COMPASS-31 secretomotor subscore. COMPASS = Composite Autonomic Symptom Scale; DLB = dementia with Lewy bodies; HR = heart rate; MoCA = Montreal Cognitive Assessment; MSA = multiple system atrophy; PD = Parkinson's disease; RBD = REM sleep behaviour disorder; SBP = systolic blood pressure; UMSARS = Unified Multiple System Atrophy Rating Scale; UPSIT = University of Pennsylvania Identification Test.
Cognition
Participants who retained the PAF phenotype had a slightly lower MoCA scores at enrolment compared to those who phenoconverted to either of the three synucleinopathies (27, IQR: 24–29 versus 30, IQR: 28–30). Lower MoCA scores (HR: 1.12, 95% CI: 1.08–1.18) and cognitive impairment (HR: 1.9, 95% CI: 1.2–3) at enrolment were associated with stable PAF at the last follow-up. Cognitive function remained stable in subjects who retained the PAF phenotype, with a median decline of one point/year (IQR: −1 to 1 point/year). In contrast, subjects that phenoconverted to DLB had the highest annual rate of decline in the MoCA (3 points/year, IQR: 2–6 points/year), compared to PD (1 point/year, IQR: 1–3 points/year) and MSA (1 point/year, IQR: 0–1 point/year).
Olfaction
At enrolment, preserved olfactory function (UPSIT score >28) was associated with a higher risk of phenoconversion to MSA (HR: 8.7, 95% CI: 1.7–45 P = 0.01). In contrast, 26 out of 30 (87%) of incident PD/DLB cases had severe hyposmia/anosmia at enrolment (Supplementary Table 2 and Supplementary Fig. 2). A decline in UPSIT scores over time was observed in patients who developed PD (4, IQR: 0–5) or DLB (2, IQR: −1– 4). Olfactory function remained stable in subjects that phenoconverted to MSA (0, IQR: −2–1) and those who remained as PAF (1, IQR: −1–2).
Autonomic predictors of phenoconversion
The strongest autonomic predictor for phenoconversion to PD was a blunting of the heart rate response to hypotension, a marker of cardiac baroreflex function. Hence, the highest hazard ratio was associated with a ΔHR/ΔSBP ratio <0.2 bpm/mmHg (HR: 6.1; 95% CI: 1.4–26) and an increase in heart <11 bpm during tilt (HR: 5.3; 95% CI: 1.1–26).
The risk of phenoconversion to any diagnosis was associated with the presence of sexual dysfunction at enrolment (UMSARS-1.11 > 0, HR: 3.6; 95% CI: 1.1–12). Subjects that reported more severe sexual dysfunction had a higher risk of developing MSA (UMSARS-1.11, HR: 2.2; 95% CI: 1.1–4.2). The presence of supine hypertension at baseline was not a strong predictor of phenoconversion, instead was associated with retention of the PAF phenotype (HR: 1.9; 95% CI: 1.2–3) (Fig. 2).
Autonomic symptom profiles
Total COMPASS-31 scores, which represent the overall burden of autonomic symptoms, were similar in phenoconverters and those who retained their PAF diagnosis. Organ-specific symptoms did, however, have a differential value. Urinary symptoms were associated with phenoconversion to any diagnosis (COMPASS-31 urinary HR: 1.3; 95% CI: 1.1–1.5). In particular, a higher burden of urinary symptoms at entry was associated with a higher relative risk of developing MSA (COMPASS-31 urinary HR: 1.6 95% CI: 1–2.5, P = 0.033). A higher score in the COMPASS-31 secretomotor score, related to changes in sweating, dry eyes and dry mouth, was also associated with a higher relative risk of phenoconversion to MSA (HR: 1.8; 95% CI: 1–3.1).
Discussion
In this prospective natural history study of patients with PAF, ∼30% of patients phenoconverted to PD, DLB or MSA after an average of 6 years of follow-up. Most phenoconversions were to Lewy body disease (PD 42% and DLB 35%), with fewer participants phenoconverting to MSA, a rare disease (23%). Average annual phenoconversion rates in patients with PAF were 12%, which is a higher yield than observed in prospective cohorts with polysomnography-confirmed RBD.4 The findings replicate the observations of our initial description of the cohort, as well as other retrospective studies, but also reveal some novel predictors, such as sexual dysfunction, symptoms of secretomotor dysfunction and specific subtle motor features. We also underscore the importance of repeated screening of cognitive function, as a single low MoCA score cannot reliably predict further cognitive impairment in this population.
Phenoconversion to any diagnosis was predicted by the presence of subtle motor signs with nOH at enrolment, including asymmetric arm swing, minimal broadness of the gait, reduced frequency of blinking, decreased facial expression or small deviations on cerebellar function tests (heal-to-shin, finger-to-nose). Autonomic markers predicting phenoconversion included sexual dysfunction and lower urinary tract symptoms. The combination of subtle motor signs, deteriorating handwriting and a more blunted orthostatic HR response to hypotension best predicted those that phenoconverted to PD from those who retained the PAF phenotype at the last follow-up.
All patients with nOH have a reduced orthostatic HR response, however, our results indicate that this reduction varies depending on the underlying neurodegenerative disorder. The HR response to OH was more severely blunted in participants who phenoconverted to PD, reflecting cardiac sympathetic noradrenergic denervation.6 This is consistent with the blunted maximum heart rate during cardiac stress testing previously reported in patients with prodromal PD.26 Thus, our study supports the inclusion of autonomic markers for predictive multi-modal modelling of phenoconversion risk.
The strongest predictors of phenoconversion to DLB were associated with self-reported subtle motor signs, such as changes in speech, dysphagia and perceived difficulties handling utensils. Many participants who retained the PAF phenotype had some degree of cognitive impairment at enrolment, but their cognitive function remained stable throughout follow-up. Our results show that a decline in cognitive scores is a predictor of imminent phenoconversion.
Urinary symptoms were important predictors of phenoconversion, especially to MSA. Because we did not perform urological examinations, we could not rule out that self-reported urinary symptoms may have been due to a urological cause (e.g. prostate hyperplasia or multiple deliveries); however neurogenic bladder is a well described feature of MSA and the fact that MSA phenoconverters had more frequent urinary symptoms at enrolment, supports the notion that these were due to autonomic, rather than non-neurological, dysfunction. Participants who phenoconverted to MSA did so at a younger age and had a younger age of PAF onset and a shorter time to phenoconversion when compared to participants who phenoconverted to DLB.
MSA accounted for the smallest number of phenoconverters, albeit still a significant percentage, considering that the incidence of MSA is at least 20 times lower than that of PD/DLB.27 Our study showed a lower proportion of PAF participants converting to MSA compared to retrospective single site cohorts.2,3,28 This can be explained by differences in study design, with our data obtained prospectively from multiple centres, which should be closer to the true population frequency. We observed a higher frequency of phenoconverters to MSA from PAF than that documented from RBD prodromal cohorts with a similar follow-up period.4,29 One possible explanation is the greater prevalence and burden of autonomic features in MSA compared to DLB and PD, leading to earlier recognition of a PAF phenotype.
Olfactory loss is a feature of patients with prodromal PD30; but was also present in patients who remained as PAF; our cohort underscores the importance of documenting preserved olfaction in patients presenting with PAF, as this is a strong predictor of phenoconversion to MSA. Our results support the use of periodic olfactory testing in patients with PAF to stratify the risk of phenoconversion as a feasible, self-administered, validated and scalable clinical biomarker for prodromal disease.11,31-33
Many participants who retained the PAF phenotype had signs of CNS involvement at enrolment, such as clinically suggestive RBD (55%), cognitive impairment (31%), olfactory loss (76%) and subtle motor signs (76%). Interestingly, most studies have shown a relatively high prevalence of RBD in patients with PAF that have not phenoconverted1,3,28 and another prospective study has also found that neither clinical RBD nor polysomnography-confirmed RBD were predictors of phenoconversion, although early-onset RBD was.3 A recent development is the use of α-synuclein seeding aggregation assays (SAA), combined with neurofilament light (NfL) chain measurement, to predict phenoconversion from PAF to MSA. Interestingly, patients who remained as PAF also showed α-synuclein SAA positivity in the CSF.34,35 Taken together, these results support the notion that, in some cases, PAF can be an indolent and slowly progressive α-synucleinopathy. It is tempting to speculate that these patients have unidentified mechanisms protecting against α-synuclein-associated neurodegeneration and slowing the spread throughout the basal ganglia and cortex. This slow disease trajectory might suggest either a neurodegenerative process caused by different α-synuclein forms, or some degree of inherent neuroprotection, promoting protein stability and/or clearance of toxic deposits and inhibiting the aggregation or spread of pathological α-synuclein throughout the CNS.
The frequency of supine hypertension at enrolment was similar in all groups by final diagnosis, but when adjusted for disease duration, supine hypertension was associated with retention of the PAF phenotype. The finding of a lower MoCA score being associated with retention of the PAF phenotype in our cohort is interesting. Comparable findings were reported in two other studies, with a similar proportion of subjects who retained the PAF phenotype with prolonged disease duration and had mild-to-moderate cognitive impairment.35,36 As in our study, the authors also noticed that phenoconverters had a normal initial cognitive evaluation. However, we did show that patients who phenoconverted to a Lewy body disease (particularly DLB), had a decline in MoCA scores at follow-up, a marker of worsening cognitive function over time.
Our study has limitations. The cohort was primarily US-based (Supplementary Table 1) and from academic medical centres that specialized in autonomic disorders, thus there may be slight differences in referral patterns in European patients, which may have an impact on how broadly representative the data were. We did not use a central laboratory to perform catecholamine analysis. Only a small subset of subjects underwent polysomnography testing to confirm their diagnosis of RBD and in most cases, clinical RBD status was determined by a questionnaire, which may not be as accurate, particularly in those who sleep without a bed partner. Despite our best efforts, lost to follow-up rates were high, making precise estimates of annual phenoconversion risk difficult and comparisons between different prospective cohorts challenging.
We are aware that RBD can precede the onset of PAF, in which case subjects would have a longer ‘disease duration’ than that defined as the onset of symptoms of nOH.
We were not able to perform subgroup analysis on patients with PAF who developed RBD before nOH, which may also be confounded by the recall abilities of the subjects.
Finally, loss to follow-up may not have been random and may have led to the omission of participants that rapidly advanced to dementia or death and were not available for repeat assessments, potentially contributing to underestimating the actual rate of phenoconversion. On the contrary, patients who were less symptomatic and more stable may have been less inclined to commit to regular follow-up.
We did not use quantitative digital technologies to assess subtle motor deficits, which may be more sensitive to detecting changes in motor function in the prodromal stages. The prodromal phase of neurodegenerative motor disorders gradually progresses until a movement disorder is evident, thus a final diagnosis depends on the assessment performed by a movement disorder specialist. As quantitative motor assessments are introduced, phenoconversion may be defined more precisely and detected earlier.
Due to the impact of COVID-19, some follow-up visits were performed via telemedicine, for which the clinical diagnostic criteria of PD, DLB and MSA have not yet been validated; this might result in an underestimation of the rate of phenoconversions. To reduce the visit burden, we assessed motor function using only the UMSARS scores, which differ slightly from the Unified Parkinson’s Disease Rating Scale, albeit both are conceptually similar. Moreover, the overall low number of patients in this cohort may have resulted in a type-II statistical error and therefore, the inability to detect other significant predictors of phenoconversion.
Although the ideal analysis of these data would use the onset of PAF symptoms as the time origin, this requires adjustment for the accompanying delayed entry (i.e. due to the requirement that the time to phenoconversion be larger than the time to study enrolment). Without this adjustment, the analyses would be biased with longer times-to-event than in the general population of subjects with PAF. However, this adjustment requires constructions of ‘risk sets’ for Kaplan Meier and Cox analyses that properly align patients according to their study enrolment and consequently reduce the effective sample size and information in the data. For this reason, we used study enrolment as the time origin in our primary analyses. As this is not a clinically meaningful origin, we adjusted for the duration of PAF symptoms at the time of study enrolment in our regression models. Ideally, future collaborations will allow for joint evaluation of multiple studies, which will enable more clinically meaningful analyses.
Despite these limitations, our main conclusion remains: a substantial group of patients with PAF have specific markers that predict their future phenoconversion to PD, DLB or MSA. Objective measurable autonomic markers could be conceivably used to calculate a prognostic risk score. For example, autonomic failure and preserved olfaction may provide a sensitive way to assemble an enriched risk cohort with prodromal MSA, the most rapidly progressive α-synucleinopathy, in whom to test potential disease-modifying therapies in relatively short clinical trials.8 The addition of fluid biomarkers, such as SAA or NfL, may strengthen the predictive ability of the identified phenoconversion markers.34
Data availability
Individual participant data (including data dictionaries) collected during the study, after de-identification might be shared with investigators of the Natural History Study of the Synucleinopathies, for any scientific purposes, after approval of a methodologically sound proposal. Proposals should be directed to the corresponding author. [email protected]
Funding
This study was funded by the National Institute of Neurological Disorders and Stroke grant U54NS065736, and supported in part by the Division of Intramural Research, National Institutes of Health, National Institute of Neurological Disorders and Stroke. The National Institute of Neurological Disorders and Stroke was involved in the study design and data collection.
Competing interests
The authors report no competing interests.
Supplementary material
Supplementary material is available at Brain online.
