Meta-analysis of Placebo-controlled Clinical Trials of Safinamide and Entacapone as Add-on Therapy to Levodopa in the Treatment of Parkinson’s Disease

The signs and symptoms of Parkinson’s disease (PD), a chronic neurodegenerative disorder predominantly characterised by the loss of dopaminergic neurons in the substantia nigra pars compacta, are most effectively treated by levodopa (L-dopa). However, nearly all PD patients experience a fluctuating response to L-dopa sooner or later within the first 5 years of therapy dependent on dosing of L-dopa. End-of-dose wearing off and dykinesias are the most common motor complications associated with L-dopa treatment.^1–3 Once they appear, the management of motor complications is often challenging and patients require recurrent drug therapy adjustments to improve fluctuations of movement without exacerbating severe dyskinesia.⁴

Safinamide recently received a European Committee for Medicinal Products for Human Use (CHMP) positive opinion for the treatment of adult patients with idiopathic PD as add-on therapy to a stable dose of L-dopa, alone or in combination with other PD compounds, in midto- late-stage fluctuating patients. It is a unique molecule with novel mechanisms of action, both dopaminergic and non-dopaminergic ones, which include monoamine oxidase-B (MAO-B) inhibition, sodium channel blockade and calcium channel modulation, thus inhibiting the excessive glutamate release. Safinamide improves motor symptoms, motor complications and quality of life in combination with other PD drugs such as dopamine agonists and L-dopa, reduces off time andextends on time without troublesome dyskinesia.^5–7 Entacapone is a potent and specific peripherally acting catechol-O-methyltransferase (COMT) with a nitro-catechol structure.⁸ It is used as an adjunct to L-dopa/dopadecarboxylase inhibitor (DDCI) therapy and slows the peripheral degradation of L-dopa only (has no anti-parkinsonian activity on its own). There are no head-to-head clinical trials comparing safinamide and entacapone under clinical conditions as an add-on therapy to L-dopa.

The aim of the present analysis is a comparison of entacapone and safinamide as add-on treatments to L-dopa in fluctuating PD patients; therefore, a meta-analysis of all the pertinent double-blind, placebo controlled studies was performed to determine effect sizes of safinamide and entacapone.

Methods
Search Strategy
The complete sets of data on placebo-controlled efficacy trials with safinamide as add-on to L-dopa were provided by Zambon, which is currently developing this still investigational compound. These comprise all completed, randomised, double-blind, placebo-controlled studies in PD patients with motor fluctuations, already treated with stable L-dopa dose and who may be receiving further anti-parkinsonian drugs (study 016, NCT01187966, labelled SAF1⁷ and SETTLE, NCT00627640, labelled SAF2^9,10,11). The duration of the trials was 24 weeks each. A systematic search of the literature up to the end of September2014 was performed on the MEDLINE and EMBASE databases using ‘entacapone’ and ‘levodopa’ or ‘L-dopa’ as search terms. Ongoing trialsand other possible completed trials still unpublished as full papers (available through clinical trial registries, conference proceedings or other literature where it proved difficult to retrieve complete data sets for the analysis) were not identified as part of the search strategy. A 2010 Cochrane Database Systematic Review on add-on therapies to L-dopa treatment was used as a primary reference for the literature.¹²

Study Selection
Eligible studies for the search were defined as any prospective, randomised, placebo-controlled and double-blinded trials (24 weeks/6 months’ duration) on the efficacy and safety of entacapone in PD patients already receiving L-dopa (usually commercial formulationsof L-dopa in a fixed combination with a DDCI), i.e. L-dopa+carbidopa, L-dopa+benserazide, with motor fluctuations. All other aspects of planned treatment were to be the same in both arms. Therefore, the following types of studies were not considered:

• Studies in healthy subjects or subjects in early-stage PD, without motor fluctuations;
• Switch studies (investigating the introduction of entacapone as add-on to L-dopa or L-dopa+DDCI), or any studies with a fixed L-dopa+DDCI+entacapone combination compared with the separate drug formulations, without parallel entacapone placebo treatment groups;
• Crossover studies – since focus was given to studies of sufficient duration to be comparable to the studies investigating safinamide;
• Studies not investigating efficacy and safety of entacapone, or entacapone and other treatments (for example, studies regarding physiology parameters, magnetic resonance imaging or pharmacokinetic parameters); and
• Studies of COMT inhibitors (COMTIs)-genotype interaction, which selected patients based on COMT gene polymorphism.

Study Appraisal and Methods
If different doses and/or dose ranges were found in the selected studies, these were to be analysed both as separate studies and collectively. Among the analysed efficacy parameters were changes from baseline in daily L-dopa dose, total daily off time, total and percent on time and Unified Parkinson’s Disease Rating Scale (UPDRS) scores (part I, II and III only). Safety was analysed mainly on binary criteria (incidence of treatment-emergent adverse events [AEs], AEs plus deaths, discontinuation from the trials and the incidence of some specific AEs). Only parameters that had been evaluated and documented for both treatments were to be included in the analyses.

The effects within studies are quantified by changes from baseline in quantitative parameters or incidences of events (non-completers, discontinuations, incidences of treatment-emergent AEs). Fixed-effect and random-effects meta-analyses are used to provide effect sizes within each treatment group and to compare both groups. Fixed-effects methods aim at estimating unknown but constant effect sizes based on the assumption that the (retrieved) studies estimate a common (fixed) effect. Random-effect methods consider the (retrieved) collection of studies as one possible random sample of a larger population, and aim at estimating the mean of a possible distribution of effect sizes.

The statistical analyses were carried out by means of Comprehensive Meta-Analysis Version 2 (CMA2) (Biostat, Englewood, NJ, US). Trial design, inclusion and exclusion criteria, blinding and randomisation of the selected trials were compared with regard to possible risk of bias. The following summary measures were used:

• Hedges’ g, which is a measure of effect size¹³ (as standardised mean difference);
• Odds ratios (ORs) (as a measure of association of binary data); and
• Risk differences (as a measure of difference between binary data).

The heterogeneity across studies was quantified by the I²-index, i.e. the ratio in percent of the true heterogeneity (the between-study variance) to the total observed variance.

Results
Study Selection
The synthesis of results was based on four studies with entacapone, and the two safinamide studies. Initially, 30 publications (published from 1996 to 2012) concerning entacapone, describing a total of 13 distinct randomised, double-blinded and placebo-controlled clinical trials, were considered. For the period up to 2008 these results matched with the findings of Stowe et al.¹² Eventually, four entacapone studies were included for the meta-analysis (see Table 1). The four studies are Poewe et al. (ENT1),¹⁴ the Parkinson Study Group (labelled ENT2),¹⁵ Rinne et al. (ENT3)¹⁶ and Brooks et al. (ENT4).¹⁷ This last trial¹⁷ recruited both fluctuating and non-fluctuating patients, but separate study results on the fluctuating patients were available; therefore, only those parts of the study results were included, whenever available (see Table 1). The mainreason for exclusion of some studies was the short duration, which did not allow appropriate comparison with the safinamide studies. All included entacapone studies considered a placebo-controlled administration of 200 mg oral dose with each dose of L-dopa. Metaanalyses were performed with all four studies whenever possible (or with only ENT1,^14,15 ENT2¹⁴ and ENT3,¹⁶ whenever outcomes for the subgroup of fluctuating PD patients were not documented in the available publications).

One of the safinamide trials (SAF1) had two dose arms (low dose and high dose, 50 mg/day and 100 mg/day, respectively), which were analysed as two separate studies (SAF1A, SAF1B, each compared with placebo) and additionally as a combined placebo-controlled dose group (SAF1C). The other safinamide trial (SAF2) had only one dose arm (50–100 mg/day), which was ascribed to high dose (hence labelled SAF2B) since after week 2 of the study period a large majority of the patients (224/274) had a dose of 100 mg/day for the rest of the study. Therefore, meta-analyses for safinamide trials are performed: (a) using all safinamide doses (SAF1A, SAF1B, SAF2B); (b) using combined safinamide doses (SAF1C, SAF2B); and (c) using the high safinamide dose (SAF1B, SAF2B) only (see Table 2).

The safinamide studies included 1,218 patients in total and the entacapone studies 849 patients. No relevant differences were detected in terms of trial design, inclusion and exclusion criteria, blinding and randomisation of the selected trials.

Efficacy Endpoints
There was a greater decrease from baseline in the daily oral L-dopa dose observed with entacapone treatment compared with treatment with safinamide all doses ( p<0.05; Figure 1 and Tables 3 and 4). This was also seen with the safinamide combined dose data, and with the random effects analyses. No significant differences between entacapone and safinamide treatment were observed in terms of total daily off time, total daily on time and percentage of on time (Figure 1 and Tables 3 and 4). This was found with safinamide all doses, with safinamide-combined doses and with the fixed- and random-effect analyses. Similarly, no statistically significant differences appeared between safinamide and entacapone treatment with respect to changes from baseline in the UPDRS total score using all safinamide doses. The alternative analysis where the two active treatment arms (low and high dose) have been combined revealed similar results to the main analysis (see Figure 1).

Safety Endpoints
Some AEs regarded as candidates for the indirect comparison of entacapone and safinamide were not considered in this report because they were not reported in the entacapone studies (anxiety, confusion and agitation, depression, dry mouth, orthostatic hypotension and syncope) or because they were reported only in entacapone trials (ataxia and forgetfulness). The analysis of the number of discontinuations due to AEs and deaths was performed after exclusion of the study ENT4 by Brooks et al.,¹⁷ because premature discontinuations in this study were not stratified by fluctuating and non-fluctuating patients.

There were no differences on study discontinuations due to AEs and deaths between safinamide and entacapone (p=0.2564), even if the number of discontinuations due to AEs occurred more frequently with entacapone treatment (OR 1.777; p=0.0464) than with safinamide (OR 1.167; p=0.5037) (Figure 2 and Tables 3 and 4). No changes in comparison with placebo were observed with safinamide treatment in terms of the number of non-completers, whereas this number was increased slightly with entacapone treatment.

A statistically significant difference in favour of safinamide was determined in the ORs versus placebo of total incidences of treatment-emergent AEs (p=0.0020) (see Tables 3 and 4). The funnel plot corresponding to the total incidences of treatment-emergent AEs identifies the trial by Poewe et al.¹⁴ as an outlier among the trials with respect to the total AE incidences (see Figure 3A). No heterogeneities were detected within both groups ofstudies however. The funnel plot generated with the alternative analysis where the two safinamide dose groups were combined confirms the result (see Figure 3B).

The risk differences of the events/symptoms versus placebo are described in Figures 4 and 5. Significant differences in favour of safinamide were seenparticularly in terms of nausea (p=0.002), vomiting (p=0.007), shortness of breath (p=0.009), urine abnormality (p=0.006), dizziness (p=0.009) and diarrhoea (p=0.001). The raw overall AE rates are 68.7 % with safinamide treatment and 84.4 % with entacapone treatment. The OR for active treatment in comparison with placebo for the total incidence of AEs were entacapone: 2.089 (p=0.0026) and safinamide: 0.907 (p=0.3880) (see Figure 6A), resulting in an indirect OR of 0.434 of safinamide versus entacapone (95 % confidence interval [CI] 0.256–0.737). The ORs for active treatment using high safinamide doses versus placebo and combined safinamide doses versus placebo are shown in Figures 6B and 6C, respectively.

Discussion
The number of trials included in this analysis is low. However, the study numbers are still valid for meta-analytical methods. In addition, the results of the present selection of entacapone studies are in line with those presented by Brooks et al.¹⁸ in a pooled analysis aimed at evaluating the efficacy and safety of long-term L-dopa/DDCI and entacapone therapy. Safinamide and entacapone treatments appeared comparable in terms of the main efficacy variables (off time, percentage on time, UPDRS). These results are in line with those obtained previously in the LARGO trial,¹⁹ comparing rasagiline to entacapone,where the efficacy of the two drugs was similar. Thus, safinamide can be considered as an efficacious drug for reducing motor fluctuations, with a magnitude similar to that of entacapone.

Advantages in favour of safinamide were detected in terms of tolerability, particularly in terms of nausea, vomiting, shortness of breath, urine abnormality, dizziness and diarrhoea. A previous meta-analysis performed by the Cochrane Collaboration,^12,20 assessing the efficacy and safety of three drug classes commonly used as adjuvant therapy to L-dopa for PD patients with motor complications (COMT inhibitors, MAO-B inhibitors and dopamine agonists), showed that COMT inhibitors reduce off time and improve UPDRS at the expense of increased onset of dyskinesia and numerous other side effects. They concluded that: “… in terms of safety, dopamine agonists and COMT inhibitors have a similar incidence of side effects, although more than MAO-B inhibitors”. In particular, the Cochrane review reported that the frequency of dyskinesia and patient withdrawal due to AEs were increased with COMT inhibitors while there was no difference between MAO-B inhibitors and placebo. These data have been confirmed by the present meta-analysis, where the number of discontinuations due to AEs, in particular dopaminergic reactions, occurred more frequently with entacapone treatment than with safinamide, despite a significant reduction in the oral L-dopa dosing with entacapone. Moreover, a retrospective pooled analysis performed by Brooks18 on four entacapone comparable trials showed that the most commonly reported AEs were aggravation of PD, dyskinesia, nausea, dizziness and diarrhoea.

Conclusions
Motor fluctuations are among the most frequent and disabling complications of L-dopa treatment. Addition of COMT inhibitors improves motor complications, but increases the risk of side effects and the complexity in administration. There is a need for new easy-to-use, efficacious treatments. Safinamide is a safe and well-tolerated drug, due to its high selectivity and reversibility, and does not have amphetamine or methamphetamine metabolites (such as selegiline).

This meta-analysis has shown that safinamide reduces disability and improves fluctuations without raising the frequency of troublesome dyskinesia with a magnitude of efficacy similar to entacapone. Moreover, safinamide was well tolerated with a safety profile not different from placebo, whereas comparison with placebo was significantly less favourable with entacapone. No dopaminergic AEs were recorded, despite a significantly lower reduction of L-dopa dose relative to entacapone, confirming that safinamide has no major drug–drug interactions and no need of diet restrictions. Furthermore, safinamide was given once a day without titration (whereas entacapone application rates vary from 3 to 10 daily in clinical practice), both relevant factors for patients usually receiving complicated oral drug regimens. In summary, these properties confirm that safinamide is straightforward to administer, is an effective and safe drug for the treatment of PD and is a favourable candidate as adjunct therapy to L-dopa and other anti-parkinsonian therapies.