SETTLE
(2017)Objective
To investigate the efficacy and safety of safinamide (oral aminoamide with dual MAO-B inhibition and sodium/calcium channel blocking) as a levodopa adjunct in patients with Parkinson disease and motor fluctuations
Study Summary
• Off time was reduced by 1.03 hours more than placebo (P<.001), and UPDRS Part III motor scores improved significantly (−1.82 vs placebo; P=.003)
• Dyskinesia was the most common adverse event with safinamide (14.6% vs 5.5%) but was rarely troublesome; overall completion rates were high (~89%)
Intervention
Safinamide 50 mg/d orally (titrated to 100 mg/d by day 14 if tolerated) vs matching placebo, once daily with breakfast for 24 weeks, as adjunct to stable levodopa and other PD medications
Inclusion Criteria
Age 30–80, idiopathic PD ≥3 years, Hoehn and Yahr 1–4 during off, daily off time >1.5 hours (excluding morning akinesia), stable levodopa regimen (3–10 doses/day) for ≥4 weeks, motor fluctuations persisting after pharmacotherapy optimization
Study Design
Arms: Safinamide 50–100 mg/d, Placebo
Patients per Arm: Safinamide 274, Placebo 275
Outcome
• Key secondary: off time reduced by 1.03 h/d more with safinamide (P<.001); UPDRS Part III improved (P=.003); CGI-C improvement 57.7% vs 41.8% (P<.001); UPDRS Part II not significant
• Quality of life (PDQ-39) improved significantly with safinamide (nominal P=.006)
Bottom Line
Safinamide 100 mg/d as a levodopa adjunct significantly increased daily on time without troublesome dyskinesia by 0.96 hour more than placebo over 24 weeks. It also significantly reduced off time (−1.03 h vs placebo), improved motor function (UPDRS Part III), CGI scores, and quality of life. The benefit was established by week 2 and maintained throughout. Dyskinesia was more common with safinamide (14.6% vs 5.5%) but was rarely troublesome, and overall tolerability was good with high completion rates (~89%). The results support safinamide as an effective once-daily oral adjunct for fluctuating PD, with effects comparable or superior to rasagiline and entacapone in cross-trial comparisons.
Major Points
- SETTLE was a 24-week, randomized, double-blind, placebo-controlled phase 3 trial of safinamide as levodopa adjunct in 549 PD patients with motor fluctuations across 119 centers in 21 countries
- Primary endpoint met: on time without troublesome dyskinesia increased by +1.42 h/d with safinamide vs +0.57 h/d with placebo (LS mean difference 0.96 h; 95% CI 0.56–1.37; P<.001)
- Key secondary endpoints: off time reduced by 1.03 h/d more than placebo (P<.001); UPDRS Part III improved by 1.82 points more (P=.003); CGI-C improvement 57.7% vs 41.8% (OR 1.92; P<.001); but UPDRS Part II was not significant (P=.15), causing hierarchical testing failure for subsequent endpoints
- PDQ-39 quality of life improved by 2.33 points more than placebo (nominal P=.006); EQ-5D improved (nominal P<.001)
- Benefit was established by week 2 on the starting dose of 50 mg/d and maintained at all subsequent assessments on 100 mg/d
- Off-time reduction of 1.03 hours compares favorably with PRESTO (0.49 h for rasagiline 0.5 mg, 0.94 h for 1 mg) and LARGO (0.78 h for rasagiline 1 mg, 0.80 h for entacapone)
- Importantly, 92% of the on-time increase was without troublesome dyskinesia (only +0.08 h increase in on time with troublesome dyskinesia, not significant)
- Dyskinesia was the most commonly reported TEAE (14.6% vs 5.5%), comparable to rasagiline in PRESTO (18% vs 10%); severe dyskinesia was uncommon (1.8% vs 0.4%)
- Only 11 patients (4.0%) on safinamide vs 21 (7.6%) on placebo required a change in antiparkinsonian treatment during the study
- Small but nominally significant worsening on UPDRS Part IV dyskinesia subscores (items 32–34 and 32–35; +0.26 difference for both; nominal P=.04)
Study Design
- Study Type
- Randomized, double-blind, placebo-controlled, parallel-group, multicenter phase 3 trial
- Randomization
- Yes
- Blinding
- Double-blind. Centralized computerized interactive voice-response system for randomization; matching tablets in matching blister packs for each dose level. Raters administering efficacy instruments were unaware of trial group assignments.
- Sample Size
- 549
- Follow-up
- 24 weeks (with 1-week taper at week 25 and 4-week follow-up safety assessment at week 29 for non-continuing patients)
- Centers
- 119
- Countries
- 21 countries across Western Europe, Eastern Europe, Asia-Pacific, and North America
Primary Outcome
Definition: Change from baseline to week 24 in mean daily on time without troublesome dyskinesia (patient diary, hours/day)
| Control | Intervention | HR/OR | P-value |
|---|---|---|---|
| +0.57 (SD 2.47) h/d from baseline 9.06 h | +1.42 (SD 2.80) h/d from baseline 9.30 h | - (0.56 to 1.37 (LS mean difference 0.96 h)) | <.001 |
Limitations & Criticisms
- Only one dose level (100 mg) was prespecified; the 50 mg dose was only a transient starting dose, so dose-response cannot be formally assessed
- 24-week duration limits long-term efficacy and safety assessment, though a prior safinamide study showed 2-year maintained benefit
- UPDRS Part II (ADL) was not significant (P=.15), causing hierarchical testing failure; all subsequent key secondary P values are nominal only
- Dyskinesia as a TEAE was nearly 3× more frequent with safinamide (14.6% vs 5.5%); while mostly non-troublesome, a small nominally significant worsening on UPDRS Part IV dyskinesia subscores was observed
- Diary-based outcomes are subject to patient reporting variability and may not capture overnight symptoms (no overnight diary data collected)
- Only days 2 and 3 of each 3-day diary period were analyzed (day 1 excluded); rationale provided but potentially reduces data
- Asian patients constituted ~31% of enrollment, but no specific ethnic subgroup analyses were presented
- Patients requiring change in PD treatment had their last pre-change assessment used for primary outcome (LOCF), which could bias results
- Open-label extension availability may have influenced patient behavior or willingness to continue in the placebo arm
- Cross-trial comparisons to rasagiline and entacapone have inherent limitations due to different populations, designs, and time periods
Citation
JAMA Neurol. 2017;74(2):216-224