A total of 208 patients with MSA (120 males, 98 MSA-C and 110 MSA-P) were included in the study (6 neuropathologically established, 174 clinically established, and 28 clinically probable). Mean age at disease onset was 57.6 ± 8.5 years, mean disease duration was 7.5 ± 3.8 years. At the time of the analysis, 157 (75.5%) were deceased. The most common causes of death were sudden death (death during sleep, respiratory failure and cardio-respiratory arrest), bronchopneumonia and urinary infection. Demographic and clinical features of the study sample are shown in Table 1.

On the total sample, 180 patients underwent VPSG for stridor and/or RBD suspicion at history taking.

Overall, 91 (47.4%) patients were diagnosed with stridor during sleep, 41 patients were identified with early stridor onset, 10 of these presented stridor as first symptom of disease. One patient showed both stridor during sleep and wakefulness. Median latency of stridor onset was 4 (2–6) years and median disease duration after stridor onset was 3 (1–5) years. Concerning stridor treatment, 28 patients were treated with tracheostomy and 36 with CPAP, while 27 patients did not receive treatment. Among patients without treatment 4 patients did not tolerate CPAP and refused tracheostomy, 11 patients refused any treatment or otorhinolaryngology visit, 1 patient died 5 days after stridor diagnosis (before CPAP titration), while 11 for unknown reasons (deceased from 1992 to 1997, missing data).

RBD diagnosis was confirmed in 160 patients (76.9% of the total sample), 127 of these were identified with early RDB.

The median of CSF NfL levels, available in 87 MSA patients, resulted of 3009 (2063–4216) pg/ml. CSF collection for NfL was performed, for each patient, during the first inpatient evaluation, with a median latency from disease onset of 4 (3–6) years.

Features of the MSA population with early and late stridor onset are compared in Table 2.

Compared to patients with late stridor onset, patients with early stridor onset showed a higher age at disease onset (59.3 ± 8.5 vs. 54.6 ± 8.5, p = 0.0114) and more frequently presented with autonomic onset (78.1% vs. 56.0%, p = 0.036). Disease duration was shorter in patients with early stridor onset (5.8 ± 2.4 vs. 9.1 ± 4.2, p = 0.0001).

No difference between the two groups was found concerning stridor treatment and latency for stridor treatment.

During the disease course, occurrence of symptomatic OH was more frequently reported in early stridor onset group than late stridor onset one (87.8% vs. 72.0%, p = 0. 048).

Patients with early stridor onset, when compared to those with late stridor onset, showed an earlier onset of pyramidal signs [3 (3–4) vs. 5 (3–6), p = 0.0019], urinary urgency/frequency [0 (0–1) vs. 2 (0–4), p = 0.0050] and urinary incontinence [2 (0–4) vs. 4 (2–8), p = 0.0047]. A shorter latency of symptomatic OH [2 (0.5–3) vs. 3 (0–5), p = 0.1846], urinary retention [1 (0–2.5) vs. 3 (0–5), p = 0.0712] and RBD [− 1 (− 3 and − 1) vs. 0 (− 2.5 and 2), p = 0.0852] was reported without reaching statistical significance (Table 2 and Fig. 1A).

Disease progression in Multiple System Atrophy patient subgroups. A Latencies of symptoms/signs onset and milestones of disease progression in patients with early and late stridor onset; B early (presenting within 3 years of disease onset) symptoms and signs in patients with early and late stridor onset; C Latencies of symptoms/signs onset and milestones of disease progression in patients with early onset of stridor and/or RBD and in patients with late onset of stridor and RBD; D early (presenting within 3 years of disease onset) symptoms and signs in patients with early onset of stridor and/or RBD and in patients with late onset of stridor and RBD. Data are expressed as median and interquartile range in years. * = p value < 0.05 (statistically significant); OH orthostatic hypotension, PEG percutaneous endoscopic gastrostomy, RBD REM sleep behavior disorder, y = years

Considering frequency of early symptoms and signs (presenting within 3 years of disease onset), patients with early stridor onset more frequently reported early urinary urgency/frequency (85.4% vs. 66.0%, p = 0.034), early urinary retention (48.8% vs. 38.0%, p = 0.039), early symptomatic OH (68.3% vs. 42.0%, p = 0.012) and early RBD (87.8% vs. 70.0%, p = 0.004) (Table 2 and Fig. 1B).

Concerning milestones of disease progression, the early stridor onset group showed a shorter latency of frequent falls [3 (2–4.5) vs. 4 (3–6), p = 0.0390], urinary catheterization [3 (1–4) vs. 5.5 (4–8), p = 0.0004], unintelligible speech [5 (4–5) vs. 7 (5–8), p = 0.0122], severe dysphagia/PEG [4 (4–5) vs. 8 (5–10), p = 0.0289] and wheelchair dependency [4 (3–5) vs. 7 (5–8), p = 0.0003], when compared to the late stridor onset group. However, the late stridor onset group more frequently reached unintelligible speech and severe dysphagia/PEG during the disease course (Table 2 and Fig. 1A).

In the stridor subgroup, MSA patients with early stridor onset (n = 15) showed higher CSF NfL levels than those with late stridor onset (n = 30), but without reaching statistical significance [3825 (2420–4903) vs. 3169.5 (2399–4448), p = 0.0980] (Table 2).

As previous results on this cohort showed a more rapid progression of disease in MSA patients with RBD predating disease onset [12], we repeated the analysis, in all sample, combining data on patients with early stridor onset and/or early RBD onset. Patients with early stridor and/or RBD onset (n = 132) were compared with those with late stridor and RBD onset (n = 76). Features of these two subgroups are shown in Table 3.

Concerning symptoms/sign latencies during the disease course, this subgroup, compared to patients with late onset of stridor and RBD, showed an earlier onset of pyramidal signs [3 (2–5) vs. 5 (3–7), p < 0.001], urinary urgency/frequency [0 (0–1) vs. 1 (0–4), p = 0.0314], urinary retention [2 (0–4) vs. 4 (0–6), p = 0.0281], urinary incontinence [3 (1–5) vs. 6 (4–8), p = 0.004] and symptomatic OH [2 (0–3) vs. 3.5 (1–6), p = 0.0025] (Table 3 and Fig. 1C).

Patients with early stridor and/or RBD onset showed less frequently parkinsonian features as mode of disease onset and more frequently reported early urinary urgency/frequency, early urinary retention, early urinary incontinence and early symptomatic OH (Table 3 and Fig. 1D).

MSA patients with early stridor and/or RBD onset demonstrated a more rapid progression with shorter latency of all milestones of disease progression: frequent falls [4 (2–5) vs. 5 (3–7), p = 0.0021], urinary catheterization [4 (2–6) vs. 6 (5–10), p = 0.0001], unintelligible speech [6 (4–7)vs. 8 (7–10), p = 0.0045], dysphagia/PEG [6 (5–8) vs. 8 (7–13), p = 0.0041], wheelchair dependency [5 (4–7) vs. 7 (6–9.5), p = 0.0004] (Table 3 and Fig. 1C).

On the whole sample, the subgroup of patients with early stridor and/or RBD onset (n = 68), showed higher CSF NfL than those with late stridor and RBD onset (n = 19) [3260 (2366–4569) vs. 1993 (1694–2766), p = 0.0015], despite latency from disease onset to CSF collection was shorter in the first subgroup than in the second one [3.5 (2–5) vs. 6 (5–7) years, p = 0.006] (Table 3).

Kaplan–Meier estimates of death in the overall population are shown in Supplementary Fig. 1A. In this analysis, the median duration of illness was 7.47 years.

The risk of death estimated by Kaplan–Meier analysis (Supplementary Fig. 1B) was higher in patients with early stridor than those with late stridor (p < 0.0001, log-rank test), with an incidence rate of death of 13.5 per 100 person-years in the first group and 9.7 per 100 person-years in the second group.

MSA patients with early RBD onset showed a higher risk of death than patients with late RBD onset (p < 0.0001, log-rank test) (Supplementary Fig. 1C), with an incidence rate of death of 9.8 per 100 person-years in the first group and 6.6 per 100 person-years in the second group.

Considering latencies of both stridor and RBD, the Kaplan–Meier analysis (Supplementary Fig. 1D) showed a higher risk of death in patients with early stridor and/or RBD onset than those with late stridor and RBD onset (p = 0.048, log-rank test).

The univariate and multivariate Cox regression analyses identified the following as factors associated with short survival: age at disease onset [HR = 1.05 (1.03–1.07), p < 0.001], autonomic disease onset [HR = 1.54 (1.10–2.14), p = 0.011], early symptomatic OH [HR = 1.96 (1.41–2.72), p < 0.001], early urinary urgency/frequency [HR = 1.66 (1.17–2.36), p = 0.004], early urinary retention [HR = 2.47 (1.52–4.02), p < 0.001], early incontinence [HR = 2.39 (1.48–3.85), p < 0.001], early stridor [HR = 2.16 (1.44–3.25), p < 0.001] and early RBD [HR = 2.98 (1.62–5.46), p < 0.001] (Table 4). In the multivariable model, early stridor onset [HR = 1.72 (1.04–2.86), p = 0.036] and early RBD onset [HR = 2.67 (1.35 – 5.27), p = 0.005] remained independent predictor of mortality after adjustment for autonomic onset and age at disease onset.