Pharmacological treatment for memory disorder in multiple sclerosis

  • Review
  • Intervention

Authors


Abstract

Background

This is an update of the Cochrane review "Pharmacologic treatment for memory disorder in multiple sclerosis" (first published in The Cochrane Library 2011, Issue 10).

Multiple sclerosis (MS) is a chronic immune-mediated, inflammatory, demyelinating, neurodegenerative disorder of the central nervous system (CNS) and can cause both neurological and neuropsychological disability. Both demyelination and axonal and neuronal loss are believed to contribute to MS-related cognitive impairment. Memory disorder is one of the most frequent cognitive dysfunctions and presents a considerable burden to people with MS and to society due to the negative impact on function. A number of pharmacological agents have been evaluated in many existing randomised controlled trials for their efficacy on memory disorder in people with MS but the results were not consistent.

Objectives

To assess the absolute and comparative efficacy, tolerability and safety of pharmacological treatments for memory disorder in adults with MS.

Search methods

We searched the Cochrane Multiple Sclerosis and Rare Diseases of the Central Nervous System Group Trials Register (24 July 2013), PsycINFO (January 1980 to 26 June 2013) and CBMdisc (1978 to 24 June 2013), and checked reference lists of identified articles, searched some relevant journals manually, registers of clinical trials and published abstracts of conference proceedings.

Selection criteria

All double-blind, randomised controlled parallel trials on pharmacological treatment versus placebo or one or more pharmacological treatments in adults with MS who had at least mild memory impairment (at 0.5 standard deviations below age- and sex-based normative data on a validated memory scale). We placed no restrictions regarding dose, route of administration and frequency; however, we only included trials with an administration duration of 12 weeks or greater.

Data collection and analysis

Two review authors independently assessed trial quality and extracted data. We discussed disagreements and resolved them by consensus among review authors. We contacted principal investigators of included studies for additional data or confirmation.

Main results

We included seven randomised controlled trials (RCTs) involving 625 people mostly with relapsing-remitting, secondary-progressive and primary-progressive MS, evaluating the absolute efficacy of donepezil, ginkgo biloba, memantine and rivastigmine versus placebo in improving memory performance with diverse assessment scales. Overall, clinical and methodological heterogeneities existed across these studies. Moreover, most of them had methodological limitations on non-specific selections of targeted sample, non-matched variables at baseline or incomplete outcome data (high attrition bias). Only the two studies on donepezil had clinical and methodological homogeneity and relatively low risks for bias. One RCT evaluating estriol versus placebo is currently ongoing.

We could not carry out a meta-analysis due to the heterogeneities across studies and the high attrition bias. A subgroup analysis for donepezil versus placebo showed no treatment effects on total recall on the Selective Reminding Test (mean difference (MD) 1.68; 95% confidence interval (CI) -2.21 to 5.58), total correct scores on the 10/36 Spatial Recall Test (MD -0.93; 95% CI -3.18 to 1.32), the Symbol Digit Modalities Test (MD -1.27; 95% CI -3.15 to 0.61) and the Paced Auditory Serial Addition Test (2+3 sec) (MD 2.23; 95% CI -1.87 to 6.33). Concerning safety, the main adverse events were: diarrhoea (risk ratio (RR) 3.88; 95% CI 1.66 to 9.05), nausea (RR 1.71; 95% CI 0.93 to 3.18) and abnormal dreams (RR 2.91; 95% CI 1.38 to 6.14). However, the results in both studies were subjected to a serious imprecision resulting from the small sample sizes and the low power of test (lower than 80%), which contributed to a moderate quality of the evidence. No serious adverse events were attributed to the treatments in all experimental groups.

Authors' conclusions

We found no convincing evidence to support the efficacy of pharmacological symptomatic treatment for MS-associated memory disorder because most of available RCTs had a limited quality. Whether pharmacological treatment is effective for memory disorder in patients with MS remains inconclusive. However, there is moderate-quality evidence that donepezil 10 mg daily was not effective in improving memory in MS patients with mild memory impairment, but had a good tolerability. Adverse events such as nausea, diarrhoea and abnormal dreams were not frequent but were associated with treatment. Ginkgo biloba, memantine and rivastigmine were safe and well tolerated and no serious adverse effects were reported. Future large-scale RCTs with higher methodological quality are needed.

Plain language summary

Pharmacological agents as symptomatic treatment for memory disorder in people with multiple sclerosis

This is an update of the Cochrane review "Pharmacologic treatment for memory disorder in multiple sclerosis" (first published in The Cochrane Library 2011, Issue 10).

Background

Memory disorder is one of the most frequent cognitive impairments (where people have difficulty with thinking, learning and memory) in people with multiple sclerosis (MS), affecting an estimated 40% to 60% of people and represents a considerable burden to those people. Data on the use of agents such as donepezil, ginkgo biloba, memantine and rivastigmine, either by mouth (orally) and by other means (parenterally), seemed to provide promising results. The authors of this review tried to assess the extent by which these treatments produced beneficial results (efficacy), whether they had side effects (tolerability) and whether they were safe in adults with all the types of MS.

Study characteristics

We searched medical databases for randomised controlled trials (where participants were randomly allocated to one of two or more treatment groups) of adults with MS. Neither patients nor researchers were told which treatment was given. Doctors had diagnosed MS and memory impairment using standard methods.

Key results and quality of evidence

Up to July 2013, we found only seven studies with 625 participants that met our requirements although the quality of the included studies was overall low. We did not find convincing evidence to support the use of these drugs as effective symptomatic treatments for memory disorder in people with MS. There was moderate-quality evidence that donepezil 10 mg daily was ineffective in improving memory in MS patients with mild memory problems, but it was well tolerated. Side effects such as nausea, diarrhoea and abnormal dreams, although not frequent, were reported for donepezil, while ginkgo biloba, memantine and rivastigmine were well tolerated and no serious side effects were reported.

There is one ongoing study that may ultimately provide valuable evidence in the future updates.

Summary of findings(Explanation)

Summary of findings for the main comparison. Donepezil for memory disorder in multiple sclerosis
  1. 1 Both studies had a small sample size and a power of test lower than 80%.

Donepezil for memory disorder in multiple sclerosis
Patient or population: Patients with memory disorder in multiple sclerosis
Settings: USA
Intervention: Donepezil
OutcomesIllustrative comparative risks* (95% CI)Relative effect
(95% CI)
No of Participants
(studies)
Quality of the evidence
(GRADE)
Comments
Assumed riskCorresponding risk
Control Donepezil
The mean change of total recall on the SRT from baseline
Scale from: 0 to 72
Follow-up: median 24 weeks
The mean change of total recall on the SRT from baseline ranged across control groups from 0.7 to 1.7The mean change of total recall on the SRT from baseline in the intervention groups was 1.68 higher (2.21 lower to 5.58 higher)-189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
The mean change of total correct score on the 10/36 SRT from baseline
Scale from: 0 to 40
Follow-up: median 24 weeks
The mean change of total correct score on the 10/36 SRT from baseline ranged across control groups from
1.2 to 16
The mean change of total correct score on the 10/36 SRT from baseline in the intervention groups was 0.93 lower (3.18 lower to 1.32 higher)-189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
The mean change of total correct score on the SDMT from baseline
Scale from: 0 to infinity
Follow-up: median 24 weeks
The mean change of total correct score on the SDMT from baseline in the control groups was 2The mean change of total correct score on the SDMT from baseline in the intervention groups was 1.27 lower (3.15 lower to 0.61 higher)-189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
The mean change of total correct score on the PASAT (2+3 sec) from baseline
Scale from: 0 to 120
Follow-up: median 24 weeks
The mean change of total correct score on the PASAT (2+3 sec) from baseline ranged across control groups from 0.8 to 3.5The mean change of total correct score on the PASAT (2+3 sec) from baseline in the intervention groups was 2.23 higher (1.87 lower to 6.33 higher)-189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
The number of patients experiencing diarrhoea
Follow-up: median 24 weeks
Low RR 3.88
(1.66 to 9.05)
189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
65 per 1000 252 per 1000
(108 to 588)
The number of patients experiencing nausea
Follow-up: median 24 weeks
Low RR 1.71
(0.93 to 3.18)
189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
140 per 1000 239 per 1000
(130 to 445)
The number of patients experiencing abnormal dreams
Follow-up: median 24 weeks
Low RR 2.91
(1.38 to 6.14)
189
(2 studies)
⊕⊕⊕⊝
moderate 1
-
86 per 1000 250 per 1000
(119 to 528)
*The basis for the assumed risk (e.g. the median control group risk across studies) is provided in footnotes. The corresponding risk (and its 95% confidence interval) is based on the assumed risk in the comparison group and the relative effect of the intervention (and its 95% CI).
10/36 SRT: 10/36 Spatial Recall Test;CI: confidence interval; PASAT: Paced Auditory Serial Addition Test;RR: risk ratio; SDMT: Symbol Digit Modalities Test; SRT: Selective Reminding Test;
GRADE Working Group grades of evidence
High quality: Further research is very unlikely to change our confidence in the estimate of effect.
Moderate quality: Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.
Low quality: Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.
Very low quality: We are very uncertain about the estimate.

Background

Description of the condition

Multiple sclerosis (MS) is an autoimmune inflammatory disorder characterised by demyelination and axonal and neuronal loss in the central nervous system (CNS). Besides the evident lesions within the white matter of the CNS, extensive damage of grey matter is also involved in the pathology of MS (Geurts 2012), and both cortical demyelination and axonal loss are believed to contribute to neurological and neuropsychological disability in MS (Vercellino 2005 ; Calabrese 2012).

(1) The patterns of multiple sclerosis-related cognitive dysfunctions

The most frequently affected cognitive domains in MS include memory, processing speed, attention and executive function (Rao 1991a). Among them, memory disorder is the most consistently impaired cognitive function and is seen in 40% to 65% of patients; furthermore, MS-related memory dysfunctions most typically affect long-term and working memory (Rao 1993). It was thought that memory difficulties in MS were due to impaired retrieval from long-term storage (Thornton 2002), and inadequate acquisition of new knowledge (Chiaravalloti 2008). Cognitive changes occur in patients with all MS subtypes, even in those presenting with a first event suggestive of MS known as clinically isolated syndrome (CIS) (Feuillet 2007 ; Potagas 2008), and tend to decline over time (Glanz 2012). Cognitive impairment differs in different disease courses (Huijbregts 2004), and different phases of the disease (Foong 1998; Morrow 2011). Patients with primary-progressive MS (PPMS) presented with a wide range of cognitive deficits in information processing speed, attention, working memory, executive function and verbal episodic memory, whereas the impairments in patients with relapsing-remitting MS (RRMS) were limited to information processing speed and working memory. Cognitive deficits were more severe in patients with PPMS than in patients with RRMS (Ruet 2013). Depression and fatigue, as two concurrent disorders, are strongly associated with perceived cognitive functioning (Kinsinger 2010). MS-related cognitive dysfunction has a great negative impact on the quality of life (QoL) of patients with MS, concerning activities of daily living, employment and social functioning (Rao 1991b).

(2) The mechanism underlying cognitive impairment in multiple sclerosis

The exact mechanism underlying cognitive impairment in MS has not been fully elucidated. However, it is considered that disconnection of cognitively important processing regions by injury to the interconnecting white matter is a potential mechanism for cognitive impairment in MS. MS-related cognitive dysfunction results from a series of domain-specific disconnection phenomena. As such, disruption of critical white-matter tracts will lead to reduced functional connectivity between cortico-cortical and cortico-subcortical cognitive processing regions, resulting in impairment to specific cognitive domains (Dineen 2009; Leocani 2000; Mesaros 2012; Rossi 2012). In addition, grey matter lesions and atrophy also contribute to the development of cognitive impairment in MS (Amato 2004; Calabrese 2009; Morgen 2006; Sanfilipo 2006). Regional lobar atrophy predicts MS-associated memory disorder (Benedict 2005), particularly the deep grey matter atrophy and the mesial temporal lobe atrophy (Benedict 2009).

(3) Cognitive tests used in neuropsychological test batteries

Two cognitive batteries are particularly reliable and validated in MS, being widely used in clinical practice and also for research purposes: the Brief Repeatable Battery of Neuropsychological tests (BRBN) (Boringa 2001) and the Minimal Assessment of Cognitive Function in MS (MACFIMS) (Benedict 2006). The BRBN, from a previous application of a comprehensive neuropsychological battery (Rao 1990a), includes the selective reminding test (SRT) for auditory-verbal memory and the 10/36 spatial recall test (10/36 SRT) for visual-spatial memory, the paced auditory serial addition test (PASAT) and the symbol digit modality test (SDMT) for processing speed/working memory, and the controlled oral word association test (COWAT) for language. The MACFIMS includes the PASAT and the SDMT, the California Verbal Learning Test-II (CVLT-II) for auditory-verbal memory, and the Brief Visuospatial Memory Test-Revised (BVMTR) for visual-spatial memory, the D-KEFS Sorting Test for executive functions, the Judgment of Line Orientation Test for Visual perception/Spatial processing and the COWAT for language.

Description of the intervention

Pharmacological interventions for memory disorder in MS have been based mainly on the use of disease-modifying drugs (DMDs) (such as interferon β-1a, interferonβ-1b, glatiramer acetate, natalizumab, mitoxantrone, fingolimod, teriflunomide, dimethyl fumarate, alemtuzumab). Furthermore, specific therapies for memory disorder include: acetylcholinesterase inhibitors (AChEI) (such as donepezil and rivastigmine), psychostimulants (methylphenidate, L-amphetamine sulphate, modafinil, amantadine and pemoline), N-methyl-D-aspartate (NMDA) antagonist (such as memantine) and some other pharmacological agents (such as Ginkgo biloba (GB) extracts, aminopyridines and cannabinoids).

How the intervention might work

The beneficial effects of DMDs may occur in the short term due to the anti-inflammatory effects of the therapy, and in the long term due to the protective effects on tissue damage in the brain. They may prevent or reduce the progression of cognitive dysfunction by containing the development of new cerebral lesions or by reducing the progression of brain atrophy (Amato 2006).

Cholinergic deficits, which can underlie the cognitive deficits seen in patients with MS, may derive from disruption of cholinergic pathways and impaired axonal transport of acetylcholine due to demyelination and axonal transection (Ruberg 1987). An additional mechanism of action may include increases in cerebral glucose utilisation (Amato 2006). Further reasons to favour AChEI in the treatment of MS patients lie in the hypothesis that several structures associated with the cholinergic system, such as the temporal lobe, hippocampus and thalamus, could be related to cognitive impairment in MS, especially memory (Paulesu 1996).

On the basis of functional imaging studies, it is hypothesised that fatigue and cognitive dysfunction may share a common physiopathological substrate, due to the reduction of glucose metabolism in the brain (Bakshi 2003), but the mechanism of the potential action of symptomatic agents for cognition remains unclear.

Glutamate toxicity has been implicated in the pathogenesis of MS (Pitt 2000). What role NMDA antagonist receptors and glutamate toxicity may play in cognitive dysfunction is uncertain.

GB extracts may improve cognitive performance through their effects on acetylcholine release (Nathan 2000), glycine and gamma-aminobutyric acid (GABA) receptors (Huang 2004). GB extracts also contain compounds that have neuroprotective effects (Ahlemeyer 2003).

L-amphetamine sulphate is a psychostimulant that is used to treat attention deficit disorder. Its major mechanisms of action are stimulation of dopamine release and dopamine-reuptake inhibition; meanwhile, it produces equivalent increases in noradrenaline (norepinephrine) in the hippocampus and cortex (Kuczenski 1995).

Why it is important to do this review

Some existing randomised controlled trials with these oral and parenteral pharmacological agents seemed to provide promising results. However, it is uncertain whether pharmacological treatments are really effective in treating memory disorder in patients with MS and which pharmacological treatment is the most efficient and least harmful. A systematic review of all relevant randomised controlled trials is the best way to resolve this uncertainty.

This is an update of the Cochrane review "Pharmacologic treatment for memory disorder in multiple sclerosis" (first published in The Cochrane Library 2011, Issue 10).

Objectives

To assess the absolute and comparative efficacy, tolerability and safety of pharmacological treatments for memory disorder in patients with MS.

Methods

Criteria for considering studies for this review

Types of studies

All double-blind, randomised controlled parallel trials on pharmacological treatment for memory disorder in patients with MS. We excluded quasi-randomised, single-blind or unblinded trials.

Types of participants

Adults with a diagnosis of definite MS according to the Poser's (Poser 1983) and McDonald's (McDonald 2001; Polman 2005; Polman 2011) diagnostic criteria, all of MS subtypes (RRMS, secondary-progressive (SPMS), PPMS or progressive-relapsing (PRMS)), who displayed at least mild memory impairment at 0.5 standard deviations below age- and sex-based normative data on a validated memory scale.

Types of interventions

Experimental intervention: pharmacological treatments for memory disorder without restrictions regarding dose, route of administration, frequency. Administration duration 12 weeks or greater.
Control intervention: placebo treatment or one or more pharmacological treatments.

Types of outcome measures

Primary outcomes

(1) Memory performance at week 12 (or later), as measured with one of the following memory scales:

  • 7/24 Spatial Recall Test (7/24 SRT) (Rao 1991c);

  • Logical Memory (LM) subset of the Wechsler Memory Scale-Revised (WMS-R) (Wechsler 1987);

  • Rey Auditory Verbal Learning Test (RAVLT) (Lezak 2004);

(2) Adverse events (AEs) at week 12 (or later): the number of patients with the five most frequent AEs reported in studies.

Secondary outcomes

(1) Clinical Global Impression of Change (CGIC) on memory at week 12 (or later).

(2) MS-specific health-related QoL at week 12 (or later). The following scales were accepted: the Medical Outcomes Study (MOS) 36-Item Short-Form Health Survey (SF-36) (Ware 1992) or the Multiple Sclerosis Quality of Life-54 (MSQOL-54) (Vickrey 1995) or the Multiple Sclerosis Quality of Life Inventory (MSQLI) (Fischer 1999) or the Functional Assessment of Multiple Sclerosis (FAMS) (Cella 1996).

Search methods for identification of studies

We applied no language restrictions to the search.

Electronic searches

The Review Group Trials Search Co-ordinator searched the Cochrane Multiple Sclerosis and Rare Diseases of the Central Nervous System Group Specialised Register (24 July 2013), which, among other sources, contains trials from:

  1. the Cochrane Central Register of Controlled Trials (CENTRAL) (Issue 3, 2013);

  2. MEDLINE (PubMed) (1966 to 24 July 2013);

  3. EMBASE (Embase.com) (1974 to 24 July 2013);

  4. CINAHL (EBSCOhost) (1981 to 24 July 2013);

  5. LILACS (Bireme) (1982 to 24 July 2013);

  6. PEDro (1990 to 24 July 2013);

  7. Clinical trials.gov (www.clinicaltrials.gov);

  8. World Health Organization (WHO) International Clinical Trials Registry Portal (apps.who.int/trialsearch/).

Information on the Cochrane Multiple Sclerosis Group's Trials Register and details of search strategies used to identify trials can be found in the 'Specialised Register' section within the Cochrane Multiple Sclerosis and Rare Diseases of the Central Nervous System Group's module.

The keywords used to search for this review are listed in Appendix 1.

In addition to the search run by the Trials Search Co-ordinator, we searched:

1. PsycINFO (1980 to 26 June 2013);
2. CBMdisc (1978 to 24 June 2013).

The search strategies are listed in Appendix 2 and Appendix 3.

Searching other resources

We checked the reference lists of identified articles and manually searched the following journals: Annals of Neurology (1994); Multiple Sclerosis (2004, 2011 and 2012) and Neurology (1999, 2001 and 2012).

We handsearched the abstracts of the 25 to 28 Congress of the European Committee for Treatment and Research in Multiple Sclerosis (ECTRIMS), the 15th annual meeting of Americas Committees for Treatment and Research in Multiple Sclerosis (ACTRIMS) (5 June 2010), the 5th Joint Triennial Congress of the European and Americas Committees for Treatment and Research in Multiple Sclerosis (ECTRIMS/ACTRIMS)(19 to 22 October 2011) and the World Congress on Treatment and Research in Multiple Sclerosis (joint meeting of ECTRIMS, ACTRIMS and LACTRIMS) (17 to 20 September 2008). We also communicated with authors of identified studies in an effort to identify further published, unpublished and ongoing studies.

Data collection and analysis

Selection of studies

Two review authors (He and Dong) independently screened titles and abstracts of the citations retrieved by the literature search for inclusion or exclusion. We obtained the available full texts of potentially relevant studies for further assessment. We independently evaluated the eligibility (on the basis of information available in the published data) of these studies. We listed papers that did not meet the inclusion criteria in the Characteristics of excluded studies table with the reasons for omission. We resolved any disagreement regarding inclusion by discussion or by referral to a third review author (Zhou) if necessary.

Data extraction and management

Two review authors (He and Dong) independently extracted data from the selected trials using standardised forms. We extracted information about study design, participants, interventions and outcome measures. We contacted principal investigators of included studies in order to request additional data or confirmation of methodological aspects of the study. We discussed and resolved disagreements by consensus among the review authors.

Assessment of risk of bias in included studies

We based the methodological criteria on the Cochrane Handbook for Systematic Review of Interventions Version 5.1.0 (Higgins 2011). Two review authors (He and Dong) independently evaluated the methodological quality of the studies using the 'Risk of bias' tool under the domains of sequence generation, allocation concealment, blinding, incomplete outcome data, selective outcome and other biases. We discussed and resolved disagreements among the review authors on the methodological quality of the identified studies by consensus.

Measures of treatment effect

We treated the data on memory scales and QoL scales as continuous outcomes because both were longer ordinal rating scales and had a reasonably large number of categories, therefore a mean difference (MD) was used if trials had used the same rating scale to assess outcome, and where different rating scales had been used, the measure of the treatment difference was expressed as the standardised mean difference (SMD). CGIC on memory belonged to shorter ordinal data (binary outcomes), along with the number of patients with AEs, the treatment effect was described using risk ratio (RR).

We extracted the pre and post mean values, the mean change from baseline, standard deviation, 95% confidence interval (CI), P value, the number of patients incurring the event and the number of patients for each treatment group of individual studies. Standard deviation was calculated from the CI or t-tests when it was not reported. The standard error of the mean change was calculated from standard deviation.

Unit of analysis issues

We only included studies with a parallel group design so participants had been randomised to either intervention or control with subsequent analysis at individual allocation level. We selected a single time point (at exit) and analysed data at this time (final values), along with change score from baseline for trials in which it was presented. In future updates, if studies with multiple intervention groups are included, we will create a single pair-wise comparison by combining groups if appropriate, or we will select one pair of interventions and exclude the others.

Dealing with missing data

This review involved four different types of pharmacological agents (donepezil, GB, memantine and rivastigmine). There were clinical and methodological heterogeneities across the included studies. A high attrition bias existed in trials for most types of agents except for donepezil. Therefore, we did not carry out a meta-analysis to evaluate the efficacy of pharmacological treatment for memory disorder in MS and only performed a subgroup analysis for donepezil. We contacted the study investigators for missing data. We used a last-observation-carried-forward imputation strategy for data not missing at random and used an intention-to-treat principle for analyses. For the data assumed to be missing at random, we only analysed the available data. We addressed the potential impact of missing data on the findings of the review in the Discussion section.

Assessment of heterogeneity

We assessed clinical heterogeneity by examining the characteristics of the studies, the similarity between the types of participants, the interventions, and the outcomes as specified in the criteria for included studies. The variability in study design and the risk of bias (methodological heterogeneity) was also evaluated. We evaluated statistical heterogeneity where the clinical and methodological heterogeneities were not obvious across the included studies. When pooling trials in meta-analyses, we calculated the I2 statistic to identify heterogeneity across studies. When the I2 was greater than 30%, there was some level of heterogeneity (Higgins 2011). If tests for heterogeneity were statistically significant, but inspection of the individual results suggested that it was still logical to combine results, we calculated the overall effects using a random-effects model.

Assessment of reporting biases

We did not assess the reporting biases because of the limited number of studies included for data synthesis. In future updates, and if more studies are included, we will examine potential publication bias using a funnel plot. For continuous outcomes, we will use standard errors as the vertical axis and MDs as the horizontal axis in funnel plots. For dichotomous outcomes, we will plot odds ratios (OR) or RRs on a logarithmic scale as the horizontal axis and will use standard errors as the vertical axis.

Data synthesis

We gave a descriptive summary of the results and combined data at exit (final values), along with change score from baseline for donepezil because clinically and methodologically homogeneous RCTs with low bias were only identified with donepezil. We conducted a subgroup analysis using Review Manager 5 (Review Manager 2012). We used a fixed-effect model because each study on donepezil was estimating exactly the same quantity. For the outcome data with a statistically significant heterogeneity (I2 > 30%), but inspection of the individual results suggested that it was still logical to combine results, we used a random-effects model instead. For the outcomes treated as continuous data (memory performance as measured with memory scales and MS-specific health-related QoL as measured with QoL scales), we selected the inverse-variance fixed-effect model method and the inverse-variance random-effects model method. For the outcomes treated as dichotomous data (the number of patients with AEs and CGIC on memory), we selected the Mantel-Haenszel fixed-effect model method and the Mantel-Haenszel random-effects model method.

Subgroup analysis and investigation of heterogeneity

We carried out a subgroup analysis based on the different types of pharmacological agents, but the updated data only permitted us to conduct an analysis for donepezil because the clinical and methodological heterogeneities or high attrition bias existed across the studies for the other pharmacological agents. In future updates and if further data become available, we plan to carry out subgroup analyses based on duration of treatment, different administration routes, dosage, the severity of memory disorder or specific MS subtype.

Sensitivity analysis

We undertook sensitivity analyses to assess the influence on results of the different treatment effects by comparing RRs with ORs and on results of the same treatment effect (MD) by comparing the mean change of total score from baseline with the mean total score at exit. In future updates, where possible, we will also conduct sensitivity analyses to assess the robustness of our review results of fixed-effect model versus random-effects model assumptions and of including trials at high risk of bias as well as the effects of analysing by intention to treat and the effect size (the MD versus SMD).

Results

Description of studies

See: Characteristics of included studies, Characteristics of excluded studies, Characteristics of studies awaiting classification and Characteristics of ongoing studies tables.

Results of the search

We found 1270 articles using the search strategies. After screening of titles and abstracts, we provisionally selected 73 articles. After obtaining the full papers of the 73 articles, we found seven studies that met the inclusion criteria (Krupp 2004; Krupp 2011; Lovera 2007; Lovera 2010; Lovera 2012; Mäurer 2013; Shaygannejad 2008). We excluded 66 studies: 18 studies were non-randomised studies (Barak 2002; Flechter 2007; Greene 2000; Huolman 2011; Iaffaldano 2012; Krause 2012; Lang 2012; Lacy 2013; Mattioli 2011a; Mattioli 2011b; Melanson 2010; Oliveri 1998; Patti 2009; Patti 2010; Portaccio 2013; Selby 1998; Zéphir 2005; Zéphir 2008); 10 studies were cross-over design (Aragona 2009; Bever 1996; Bruce 2012; Cohen 1989; Decoo 2004; Rorie 2001; Rossini 2011; Sailer 2000; Smits 1994; Villoslada 2009); two studies were open-label trials (Krupp 1999; Schwid 2007); one study was an observational study with an open-label retrospective analysis (Schröder 2011); one study was a cross-sectional study (Honarmand 2011); in nine studies, administration duration was less than 12 weeks (Benedict 2008; Geisler 1996; Harel 2009; Lange 2009; Möller 2011; Morrow 2009; Morrow 2013; Uttner 2005; Wade 2004); in three studies, no memory scales were used in the outcomes (Barak 1999; Jeffery 2011; Selmaj 2012); in 18 studies, the participants did not meet the inclusion criteria for memory impairment prespecified in the protocol (Bosca 2004; Cabrera-Gomez 2003; Cohen 2002; Cohen 2010; Comi 2012; Fischer 1994; Fischer 2000; Havrdova 2006; Kappos 2010; Miller 2011; Montalban 2009; Panitch 2004; Pliskin 1994; Pliskin 1996; Rudick 2006; Weinstein 1999; Weinstock-Guttman 2012a; Weinstock-Guttman 2012b); two papers (Christodoulou 2006; Sumowski 2011), were respectively based primarily on studies previously published (Krupp 2004; Morrow 2009); and two studies were not double-blind (Wilken 2004; Wilken 2008). See Figure 1. The trial that was previously classified in ongoing studies (NCT01074619), was completed in November 2011, but we could not find a publication. In this update, we listed it in the Characteristics of studies awaiting classification table. One double-blind, randomised, placebo-controlled, parallel group study is currently ongoing (NCT01466114), and will be assessed when it is published and the results become available.

Figure 1.

Study flow diagram.

Included studies

We included seven studies. Krupp 2004 and Krupp 2011 evaluated the effect of donepezil in treating memory and cognitive dysfunction in MS patients (n = 69 and n = 120, respectively). Lovera 2007 and Lovera 2012 determined if GB improves the cognitive function in people with MS (n = 43 and n = 121, respectively). Lovera 2010 determined whether memantine improves cognitive performance among people with MS and cognitive impairment (n = 126). Shaygannejad 2008 and Mäurer 2013 evaluated the effect of rivastigmine in treating memory and cognitive dysfunction in MS patients (n = 60 and n = 86, respectively).

Characteristics of the study design

Seven studies were randomised, double-blind, placebo-controlled clinical trials with parallel design.

Characteristics of the participants

All participants had the diagnosis of definite MS according to Poser's or McDonald's diagnostic criteria (McDonald 2001; Poser 1983; Polman 2005), with an age ranging of 18 to 65 years with the exception of Shaygannejad 2008 (actual age range, 16 to 54 years). All patients exhibited stable neurological functioning for at least one month prior to study entry (Krupp 2004; Lovera 2012; Shaygannejad 2008), or without a significant MS exacerbation (Lovera 2007; Lovera 2010; Mäurer 2013), or without steroid treatment within four weeks of screening (Krupp 2011). All trials reported a baseline comparability of the characteristics of participants between treatment groups, and most of characteristics of patients at baseline were similar except for gender (Krupp 2004; Lovera 2012), mean Expanded Disability Status Score (EDSS) (Krupp 2004), MS subtypes (Krupp 2004; Lovera 2010), the z-scores for the Long Delay Free Recall (LDFR) on the CVLT-II (Lovera 2010), the subtests of digit span and visual reproduction in WMS (Shaygannejad 2008), education and disease duration (Krupp 2011), the reading scores on the Wide Range Achievement Test 3 and self report on the Occupational Functioning Questionnaire (OFQ) (Krupp 2011). No information at baseline was reported on disease-modifying therapies (Lovera 2007; Lovera 2010; Mäurer 2013), depression (Shaygannejad 2008), and fatigue (Lovera 2007; Shaygannejad 2008). Krupp 2004; Krupp 2011; Lovera 2007; and Shaygannejad 2008 reported the proportion of patients using concomitant antidepressants at baseline. Most patients displayed at least mild memory impairment at least 0.5 standard deviations below age- and sex-based normative data on a validated memory scale; however, selection of targeted sample was not fully specific for memory impairment in Lovera 2012 and Mäurer 2013. Patients in Krupp 2004 and Lovera 2007 had no severe cognitive impairment. Krupp 2004; Krupp 2011; Lovera 2007; Lovera 2010; Lovera 2012; and Mäurer 2013 excluded patients with major depression. Mäurer 2013 included patients with RRMS, SPMS and CIS. Lovera 2012 included patients with all types of MS. The other studies included patients with RRMS, SPMS and PPMS.

Characteristics of the interventions

Krupp 2004 and Krupp 2011 compared donepezil versus placebo with a 24-week treatment duration and initial dose of 5 mg per day, increasing to 10 mg per day at week four. Lovera 2007 and Lovera 2012 compared GB (120 mg twice a day) versus placebo with a 12-week treatment duration. Lovera 2010 compared memantine (10 mg twice a day (four-week titration followed by 12 weeks on the highest tolerated dose)) versus placebo with a 16-week treatment duration. Shaygannejad 2008 compared rivastigmine (1.5 mg once daily increment over four weeks to 3 mg twice daily for a total of 12 weeks) versus placebo with a 12-week treatment duration. Mäurer 2013 compared rivastigmine patches (initially 5 cm2 (4.6 mg/day), and increased to 10 cm2 (9.5 mg/day) at week four) versus placebo with a 16-week double-blind treatment duration.

Characteristics of the outcome measures

Our primary outcomes were reported in all trials, memory performance on the CVLT-II (Lovera 2007; Lovera 2010; Lovera 2012); SRT (Krupp 2004; Krupp 2011; Mäurer 2013); WMS-III (Shaygannejad 2008); 10/36 SRT (Krupp 2004; Krupp 2011; Mäurer 2013); Backward Digit Span (Shaygannejad 2008); the LM subset of the WMS-R (Shaygannejad 2008); PASAT (Krupp 2004; Lovera 2007; Krupp 2011; Lovera 2010; Lovera 2012; Mäurer 2013); and SDMT (Krupp 2004; Krupp 2011; Lovera 2007; Lovera 2010; Mäurer 2013). The BVMTR, 7/24 SRT and RAVLT were not reported in the included studies. All trials reported the frequency of AEs.

Our secondary outcomes were reported in some trials: CGIC on memory (Krupp 2011); MSQLI (Lovera 2007) and SF-36 (Lovera 2010). The MSQOL-54 and FAMS were not reported in the included studies.

Excluded studies

All of the studies that were excluded from this review and the reasons for their exclusion are available in the Characteristics of excluded studies table.

Risk of bias in included studies

Further details of this assessment are available in the relevant section of the 'Characteristics of included studies' table and are also presented in the 'Risk of bias' graph (Figure 2) and 'Risk of bias' summary (Figure 3).

Figure 2.

Risk of bias graph: review authors' judgements about each risk of bias item presented as percentages across all included studies.

Figure 3.

Risk of bias summary: review authors' judgements about each risk of bias item for each included study.

Allocation

Sequence generation and allocation concealment were adequate in all studies. The risks for selection bias were low. The randomisation lists were generated by an independent pharmacist or staff using a random number generator (Krupp 2004; Krupp 2011; Lovera 2007), a computer program that differed from a random number generator (Shaygannejad 2008), by an external clinical research organisation (CRO) using a validated system that automated the random assignment of the treatment arms (Mäurer 2013), or by an independent biostatistician or research pharmacist using an EXCEL random number generator (Lovera 2010; Lovera 2012). The pharmacist was informed of all randomisation assignments and was responsible for labelling the study drug and maintaining a master list linking the participants and their treatment assignments and the masking of the active and placebo treatments was preserved by creating treatments that looked identical (Krupp 2004; Krupp 2011; Shaygannejad 2008). In Lovera 2007, the staff at the Oregon Health and Science University Medical Informatics and Clinical Epidemiology Department generated the allocation sequence, and sequentially numbered the containers provided by the manufacturer accordingly. The remainder of the study personnel did not have access to the allocation sequence until the study was completed. In Lovera 2010, the independent statistician's staff received a fax with the participant's study ID, randomised the participant, and then sent a fax to the pharmacy indicating the treatment allocation. Only the independent statistician's staff and the research pharmacists had access to the allocation sequence until the analysis was completed. In Lovera 2012, the research pharmacists assigned the participants to either GB or matching placebo tablets and dispensed the tablets to them in identical containers. Only the research pharmacists had access to the allocation sequence. In Mäurer 2013, the investigator enrolled and assigned the participants to treatment according to the randomisation list.

Blinding

All trials were double-blinded and had low risks for performance bias and detection bias. In Krupp 2004, all clinical staff and patients were masked regarding treatment assignment. The masking of the active and placebo treatments was preserved by creating treatments that looked identical. In Lovera 2007, the blinding was implemented by using identical containers, instructions and labels. The participants, physicians and research assistants did not know the treatment allocation at any point until the data analysis was completed. In Lovera 2010, only the independent statistician's staff and the research pharmacists had access to the allocation sequence until the analysis was completed. The questionnaires administered at the end of the study indicated that blinding of the participants, evaluating physicians and evaluating research assistants was successful. In Krupp 2011, except for the pharmacists, all research staff and participants were masked regarding treatment assignment. Masking of active and placebo treatments was preserved by creating capsules that appeared identical. In Lovera 2012, only the research pharmacists had access to the allocation sequence. The research pharmacists assigned the participants to either GB or matching placebo tablets and dispensed the tablets to them in identical containers. One research assistant assessed AEs and another administered the neuropsychological tests. They were instructed not to discuss AEs between them to ensure that the cognitive test assessments would be masked. In Mäurer 2013, patients, investigator staff, people performing the assessments and data analysts remained blinded throughout the entire study period. Study drugs were identical in packaging, labelling, schedule of administration, appearance and odour. In Shaygannejad 2008, the trial was double-blinded in that both patient and physician who assessed the outcome were not aware of treatment. The masking of the active and placebo treatments was preserved by creating treatments that looked identical.

Incomplete outcome data

Most studies provided enough details about the number of and the reasons for drop-out except for Krupp 2011, where the reasons for drop-out were not recorded, however, the dropout rate and the amount of missing data were low, a last-observation-carried forward imputation strategy was used for missing data and the intention-to-treat principle was used for analyses. The risk for attrition bias was low. In Krupp 2004, the missing data and reasons were balanced between groups, the risk for attrition bias was low. In Lovera 2007, although the dropout rate and the amount of missing data were low, the reasons differed across groups. Moreover, an intention-to-treat principle was not used for analyses. The risk for attrition bias was high. In Lovera 2010, the dropout rate and the amount of missing data were high. Moreover, the reasons for drop-out differed across groups. The researchers excluded the patients who were lost to follow-up and did not use the intention-to-treat principle for analyses. The risk for attrition bias was high. In Lovera 2012, the reasons for drop-out were carefully recorded and differed across groups; however, the dropout rate and the amount of missing data were low, and the modified intention-to-treat principle was used for analyses. The risk for attrition bias was low. In Mäurer 2013, although the modified intention-to-treat principle was used for analyses, attrition bias was high. The high rate of drop-outs had potential impacts on the results. Generally, the potential impact of missing continuous outcomes increases with the proportion of participants with missing data. In Shaygannejad 2008, no participants were lost to follow-up, and the risk for attrition bias was low.

Selective reporting

All listed outcomes were adequately reported in most studies except for Lovera 2007 and Lovera 2010, in which some of outcomes were lack of data, resulting in unclear risks for reporting bias.

Other potential sources of bias

All studies had unclear risks for other bias. In Krupp 2004, the possibility of detection bias may have existed because the evaluating physician was the same as the treating physician, and biased results may have been induced due to different baseline on sex, mean EDSS and MS subtypes, even a higher percentage of donepezil patients correctly guessed their medication status. In Lovera 2007, biased results may have been induced because of a lower percentage of GB patients guessed their treatment allocation correctly and no information reported on disease-modifying therapies. In Lovera 2010, the imbalances on type of MS and the z-scores for the LDFR on the CVLT-II between two groups at baseline may have resulted in biased results. In Krupp 2011, biased results may have been induced because the donepezil group at baseline had more years of education, higher reading scores on the Wide Range Achievement Test 3 and longer self reported disease duration, and a higher percentage of donepezil patients correctly guessed their medication status. In Lovera 2012, biased results may have resulted from the difference in gender between two groups and the unspecific election of targeted sample for memory impairment due to the use of the COWAT, which is used to measure language. In Mäurer 2013, biased results may have been induced because the study was supported by Novartis Pharma GmbH, and the selection of targeted sample is not specific for memory impairment due to the use of a Multiple Sclerosis Inventarium Cognition (MUSIC). In Shaygannejad 2008, biased results may have been induced due to imbalances on the subtests of digit span and visual reproduction in WMS between two groups at baseline and no information reported on depression.

Effects of interventions

See: Summary of findings for the main comparison Donepezil for memory disorder in multiple sclerosis

Primary outcomes

Overall, clinical and methodological heterogeneities existed across the included studies. Moreover, some of studies had a high attrition bias. Such a condition did not permit us to conduct a meta-analysis to evaluate the efficacy of pharmacological treatments for memory disorder in MS. However, we performed a subgroup analysis focusing on donepezil versus placebo from Krupp 2004 and Krupp 2011. The results showed no treatment effects on total recall on the SRT (MD 1.68, 95% CI -2.21 to 5.58 (Analysis 1.1) and MD 1.23, 95% CI -3.07 to 5.53 (Analysis 1.2)), total correct scores on the 10/36 SRT (MD -0.93, 95% CI -3.18 to 1.32 (Analysis 1.3) and MD 0.29, 95% CI -2.64 to 3.23 (Analysis 1.4)), the SDMT (MD -1.27, 95% CI -3.15 to 0.61 (Analysis 1.5) and MD 1.15, 95% CI -2.71 to 5.01 (Analysis 1.6 )), and the PASAT (2+3 sec) (MD 2.23, 95% CI -1.87 to 6.33 (Analysis 1.7) and MD 5.41, -1.42 to 12.23 (Analysis 1.8)). Concerning the safety, the treatment was associated with AEs such as diarrhoea (RR 3.88, 95% CI 1.66 to 9.05 (Analysis 1.9 ) and OR 4.84, 95% CI 1.87 to 12.51 (Analysis 1.10)), nausea (RR 1.71, 95% CI 0.93 to 3.18 (Analysis 1.11) and OR 1.94, 95% CI 0.92 to 4.12 (Analysis 1.12)), abnormal dreams (RR 2.91, 95% CI 1.38 to 6.14 (Analysis 1.13) and OR 3.55, 95% CI 1.50 to 8.37 (Analysis 1.14)). Sensitivity analyses by comparing RR with OR and by comparing the mean change of total score from baseline with the mean total score at exit did not show significant influence on results of the treatment effects. We graded the quality of evidence for the trials on donepezil using the GRADE approach (Schünemann 2009). Both studies had a desirable directness regarding comparison and the target population, intervention, comparator and outcomes. No serious limitations in design and implementation existed in either studies. A statistical heterogeneity of results on the primary outcome existed between the two studies, but to a large degree, it could be explained by the minor differences of the population on age, gender, education, MS subtype, disability or severity of memory impairment. Therefore, the inconsistency was negligible. However, the results in both studies were subjected to a serious imprecision resulted from the small sample sizes and the power of test (lower than 80%), which contributed to a moderate quality of the evidence (Summary of findings for the main comparison).

Lovera 2007 reported there were no any significant differences among patients receiving GB versus placebo on the LDFR from the CVLT-II, PASAT and SDMT (exit group difference: -1.2, 95% CI -3.2 to 0.8 with CVLT-II, 1.1, 95% CI -3.1 to 4.2 with PASAT, -0.8, 95% CI -3.0 to 4.6 with SDMT), and only one person in the GB group had an AE (constipation) classified as remotely related to treatment.

Lovera 2010 reported the differences between the placebo group and memantine group in change on the LDFR from the CVLT-II and PASAT were not statistically significant (exit-baseline) (MD -0.6, 95% CI -2.0 to 0.8, P value = 0.4 with CVLT-II, MD -0.0, 95% CI -3.3 to 3.4, P value = 0.9 with PASAT), and there were no statistically significant differences between the groups on the frequency of individual AEs (bladder infection, fatigue, cough, somnolence, ataxia, spasticity, constipation and rash).

Shaygannejad 2008 reported the mean general memory score on the WMS, the LM and digit span at the end of trial did not change between rivastigmine and placebo group (exit group difference: 0.4, 95% CI -2.0 to 2.8 with general memory; 1.2, 95% CI -0.7 to 1.7 with LM, -1.3, 95% CI -1.7 to -0.9 with digit span), and reported the most common AEs associated with rivastigmine were nausea (six people), dyspepsia (five people), anorexia (five people), dizziness (two people), headache (two people), tremor (one person), anxiety (three people) and confusion (one person), appeared at the starting dose and at every increase of the dosage, but disappeared after the first month of treatment.

Lovera 2012 reported the differences (GB-placebo) at exit in the z-scores for PASAT and LDFR on the CVLT-II were not statistically significant (exit group difference: -0.2, 95% CI -0.5 to 0.1 with PASAT, 0.0, 95% CI -0.3 to 0.3 with CVLT-II). There were no significant differences in AEs. Two serious AEs occurred in the GB group (myocardial infarction and hospitalisation for depression), which were not considered to be related to treatment.

Mäurer 2013 reported patients who received rivastigmine showed a non-significant increase in total recall score on the SRT over placebo (exit-baseline) (95% CI -5.8 to 1.6; P value = 0.2576). The most frequent AE was erythema, occurring equally in both groups (15.6% in rivastigmine group and 14.6% in placebo group); the other frequent AEs included rash, nausea, nasopharyngitis, MS relapse and depression. The percentages of AEs requiring dose adjustment (3.2% and 12.1%, respectively), temporal or permanent interruption of treatment (14.9% and 19.7%, respectively) and concomitant therapy (29.8% and 40.9%, respectively) were lower in the rivastigmine group than in the placebo group. The percentage of serious adverse events was higher in the placebo group (9.1%) than in the rivastigmine group (3.2%).

See: Table 1 and Figure 4.

Figure 4.

Efficacy on the primary outcomes reported in each study.

Table 1. Data of the primary outcomes reported in each study
  1. CI: confidence interval; CVLT-II: California Verbal Learning Test II; GB: Ginkgo biloba; LDFR: Long Delay Free Recall; LS: least square; PASAT: Paced Auditory Serial Addition Test; SD: standard deviation; SDMT: Symbol-Digit Modalities Test; SEM: standard error of the mean; SRT: Selective Reminding Test; WMS: Wechsler Memory Scale.

StudyPrimary outcomesGroupsNumberPre mean values at baselinePost mean values at exit

The mean

change from

baseline

Mean difference

(95% CI) between groups

P valueNotes
Krupp 2004Total recall on SRTDonepezil3542.3 ± 9.046.8 ± 9.34.6 ± 9.13.9 (0.13 to 7.66)*0.043

Mean ± SD

*exit-baseline difference

Placebo3442.7 ± 8.843.2 ± 9.20.7 ± 6.3
Krupp 2011Total recall on SRTDonepezil61

38.0 ± 8.7

(35.7 to 40.2)

39.6 ± 8.7

(37.4 to 41.8)

1.6 ± 7.5

(0.3 to 3.6)

0.54*-

Mean ± SD (95% CI)

*exit-baseline difference

Placebo59

36.0 ± 9.8

(33.5 to 38.6)

37.7 ± 11.5

(34.7 to 40.7)

1.7 ± 7.2

(0.2 to 3.6)

-
Lovera 2007LDFR on CVLT-IIGB209.4 ± 1.08.9 ± 0.7--1.2 (-3.2 to 0.8)*-

Mean ± SEM

*Exit group difference

Placebo1910.2 ± 0.710.1 ± 0.7-
SDMTGB2047.7 ± 2.843.4 ± 1.3--0.8 (-3.0 to 4.6)*-
Placebo1945.4 ± 2.844.1 ± 1.3-
PASATGB2042.4 ± 3.148.9 ± 1.1-1.1 (-3.1 to 4.2)*-
Placebo1943.9 ± 3.047.8 ± 1.1-
Lovera 2010LDFR on CVLT-IIMemantine54--

0.9

(-0.3 to 2)

-0.6 (-2 to 0.8)*0.4

Mean change (95% CI)

*Exit-baseline difference

Placebo65--

0.2

(-0.9 to 1.3)

PASATMemantine54--

3.6

(0.7 to 6.4)

-0.0 (-3.3 to 3.4)*0.9
Placebo65--

3.6

(0.9 to 6.3)

Lovera 2012LDFR on CVLT-IIGB61-0.5 ± 1.2-0.4 ± 1.2-0.0 (-0.3 to 0.3)*-

z-scores, mean ± SD

*Exit group difference, mean (95% CI)

Placebo59-0.5 ± 1.2-0.4 ± 1.2-
PASATGB61-1.4 ± 0.8-1.3 ± 0.9--0.2 (-0.5 to 0.1)*-
Placebo59-1.2 ± 0.9-1.0 ± 1.1-
Mäurer 2013Total recall on SRTRivastigmine4343.93 ± 10.67-1.35*(-5.8 to 1.6)0.2576**Exit-baseline difference of adjusted LS-means
Placebo3848.21 ± 10.32--0.76*
Shaygannejad 2008WMSRivastigmine3060.0 ± 4.264.9 ± 5.3

4.9

(3.8 to 6.0)

0.4 (-0.2 to 2.8)*-

Mean ± SD

*Exit group difference

Placebo3060.5 ± 4.964.5 ± 3.7

4.0

(3.0 to 5.0)

Logical MemoryRivastigmine3016.1 ± 1.818.0 ± 2.0

1.9

(1.4 to 2.4)

1.6 (0.4 to 2.8)*-
Placebo3015.2 ± 2.516.4 ± 2.5

1.2

(0.7 to 1.7)

Digit SpanRivastigmine304.6 ± 0.74.6 ± 0.7

0.0

(-0.3 to 0.3)

-1.3 (-1.7 to -0.9)*-
Placebo305.5 ± 1.05.9 ± 0.9

0.4

(0.2 to 0.5)

Secondary outcomes

Lovera 2007 reported no statistically significant difference between the groups on the total score of the Perceived Deficits Questionnaire (PDQ) (MD 0, 95% CI -3.8 to 3.7) in the MSQLI and in the other QoL scales from the MSQLI, except for a significant difference between the groups of 1.5 points (95% CI 2.6 to 0.3, P value = 0.015) in the Retrospective Memory Scale of the PDQ, favouring the GB group.

Lovera 2010reported there was no significant difference between the two groups on any of the scales of the SF-36.

Krupp 2011 reported a trend was noted for the clinician's impression of memory change in favour of donepezil (37.7%) versus placebo (23.7%) (P value = 0.097).

Discussion

Summary of main results

We excluded most retrieved studies due to cross-over design or lack of randomisation or double-blinding. We excluded many RCTs meeting the pre-specification of criteria for types of studies in this review due to inadequate selection of the target population or short administration duration and follow-up. Studies regarding disease-modifying therapy (DMT) and MS-related memory disorder were not designed to assess detailed cognitive outcomes. In addition, most cognitive investigations were secondary analyses or post hoc analyses of RCTs in which the primary purpose was to reduce relapse rate and preserve overall neurological function. We included seven studies in this review, which evaluated donepezil (two studies, n = 189), GB (two studies, n = 164), memantine (one study, n = 126) and rivastigmine (two studies, n = 146) on memory disorder in MS. One trial previously classified as an ongoing study was completed in November 2011, but had not been published by June 2013; we have listed this in the studies awaiting classification. There is one ongoing study that is likely to be of relevance to this review and may ultimately provide valuable evidence in future updates.

The updated data did not permit us to carry out a meta-analysis to evaluate the efficacy of pharmacological treatments for memory disorder in MS because of the clinical and methodological heterogeneities across studies and the high attrition bias. We only conducted a subgroup analysis for donepezil versus placebo. The results showed no treatment effects on total recall on the SRT (MD 1.68, 95% CI -2.21 to 5.58), total correct scores on the 10/36 SRT (MD -0.93, 95% CI -3.18 to 1.32), SDMT (MD -1.27, 95% CI -3.15 to 0.61), PASAT (2+3 sec) (MD 2.23, 95% CI -1.87 to 6.33) and the number of patients experiencing diarrhoea (RR 3.88, 95% CI 1.66 to 9.05), nausea (RR 1.71, 95% CI 0.93 to 3.18) and abnormal dreams (RR 2.91, 95% CI 1.38 to 6.14). The quality of evidence was moderate. Sensitivity analyses by comparing RR with OR and by comparing the mean change of total score from baseline with the mean total score at exit did not show significant influence on results of the treatment effects.

Overall completeness and applicability of evidence

In this review, it was difficult to reach valid conclusions that pharmacological treatment was effective for memory disorder in MS. We are unable to give any recommendations for clinical practice of any pharmacological agent by means of DMT or symptomatic treatment for MS-associated memory disorder. Moreover, not all primary outcomes, secondary outcomes, subgroup analysis or sensitivity analyses could be performed as planned because most included RCTs had relatively poor methodological quality and had clinical and methodological heterogeneities. The results based on a subgroup analysis for donepezil versus placebo were not in favour of donepezil as a symptomatic treatment for mild memory disorder in MS. Generally, the treatment for cognitive impairment needs an adequate administration duration and follow-up. A minimum duration of administration of 12 weeks, pre-defined in the criteria of types of interventions, was a reasonable treatment length that partly avoided the inclusion of misleading evidence.

Quality of the evidence

We included seven RCTs in this review, involving 625 people and four different types of pharmacological agents to mainly evaluate the absolute efficacy in improving memory performance with diverse assessment scales by direct comparison with placebo. Overall, clinical and methodological heterogeneities existed across these studies. Furthermore, most studies had methodological limitations on non-specific selections of targeted sample, non-matched variables or incomplete outcome data (high attrition bias). Therefore, we did not carry out a meta-analysis to evaluate the efficacy of pharmacological treatments for memory disorder in MS. Whether pharmacological treatment is effective in memory disorder in patients with MS remains inconclusive. However, we performed a subgroup analysis focusing on the two studies on donepezil and graded the quality of evidence by the GRADE approach. The results in both studies were subjected to a serious imprecision resulting from the small sample sizes and the low power of test (lower than 80%), which contributed to a moderate quality of the evidence.

Potential biases in the review process

An extensive and comprehensive search was undertaken to limit bias in the review process; however, we retrieved a low number of RCTs with parallel design. The two review authors' independent assessments of eligibility of studies for inclusion in this review and the extraction of data minimised the potential for additional bias beyond that detailed in the 'Risk of bias' tables. The low rate of drop-outs and amount of missing data on the primary outcome did not have a large impact on the primary analysis. Moreover, a last-observation-carried-forward imputation strategy was used for missing data and an intention-to-treat analysis principle was used for analyses in this meta-analysis. No conflict of interests were found in relation to the review authors of this review.

Agreements and disagreements with other studies or reviews

We have found no other systematic review on pharmacological treatment for memory disorder in MS.

Authors' conclusions

Implications for practice

We found no convincing evidence from this review to support the efficacy of pharmacological symptomatic treatments for multiple sclerosis (MS)-associated memory disorder. Whether pharmacological treatment is effective in memory disorder in patients with MS remains inconclusive. There is moderate-quality evidence that donepezil 10 mg daily was not effective in improving memory in MS patients with mild memory impairment, but had a good tolerability and safety as adverse events such as nausea, diarrhoea and abnormal dreams were not frequent. Ginkgo biloba, memantine and rivastigmine were safe and well tolerated and no serious adverse effects were reported.

Implications for research

The quality of the included studies was overall low, some important variables were not matched between groups, the samples of patients were relatively small and the follow-up was short. Future RCTs with larger size are required, adequate selection of the target population and better matching on important variable (such as age, gender, education level, MS subtypes, disease duration, depression, fatigue, anxiety, concomitant medications) between groups are key to studies. MS-specific health-related quality of life should be addressed in future research.

Acknowledgements

We wish to thank Andrea Fittipaldo, Trials Search Co-ordinator and Liliana Coco, Managing Editor of the Cochrane Multiple Sclerosis and Rare Diseases of the Central Nervous System Group for their help and support in developing this review. We thank all peer reviewers, and Loredana La Mantia, the quality advisor of the Cochrane Multiple Sclerosis and Rare Diseases of the Central Nervous System Group, for their constructive comments and suggestions in this review. We also would like to thank Ms Sarah Dawson (Trials Search Co-ordinator) at the Cochrane Depression, Anxiety and Neurosis Group for assistance with the search strategy for PsycINFO. In addition, we thank Dr. Guo Duan, Dr. Hao Zilong and Dr. Wu Bo, the previous authors of this review.

Data and analyses

Download statistical data

Comparison 1. Subgroup analysis for donepezil versus placebo
Outcome or subgroup titleNo. of studiesNo. of participantsStatistical methodEffect size
1 The mean change of total recall on the Selective Reminding Test (SRT) from baseline2189Mean Difference (IV, Random, 95% CI)1.68 [-2.21, 5.58]
2 The mean total recall on the Selective Reminding Test (SRT) at exit2189Mean Difference (IV, Random, 95% CI)1.23 [-3.07, 5.53]
3 The mean change of total correct score on the 10/36 Spatial Recall Test (10/36 SRT) from baseline2189Mean Difference (IV, Random, 95% CI)-0.93 [-3.18, 1.32]
4 The mean total correct score on the 10/36 Spatial Recall Test (10/36 SRT) at exit2189Mean Difference (IV, Random, 95% CI)0.29 [-2.64, 3.23]
5 The mean change of total correct score on the Symbol Digit Modalities Test (SDMT) from baseline2189Mean Difference (IV, Fixed, 95% CI)-1.27 [-3.15, 0.61]
6 The mean total correct score on the Symbol Digit Modalities Test (SDMT) at exit2189Mean Difference (IV, Fixed, 95% CI)1.15 [-2.71, 5.01]
7 The mean change of total correct score on the Paced Auditory Serial Addition Test (PASAT) (2+3 sec) from baseline2189Mean Difference (IV, Random, 95% CI)2.23 [-1.87, 6.33]
8 The mean total correct score on the Paced Auditory Serial Addition Test (PASAT) (2+3 sec) at exit2189Mean Difference (IV, Fixed, 95% CI)5.41 [-1.42, 12.23]
9 The number of patients experiencing diarrhoea2189Risk Ratio (M-H, Fixed, 95% CI)3.88 [1.66, 9.05]
10 The number of patients experiencing diarrhoea2189Odds Ratio (M-H, Fixed, 95% CI)4.84 [1.87, 12.51]
11 The number of patients experiencing nausea2189Risk Ratio (M-H, Fixed, 95% CI)1.71 [0.93, 3.18]
12 The number of patients experiencing nausea2189Odds Ratio (M-H, Fixed, 95% CI)1.94 [0.92, 4.12]
13 The number of patients experiencing abnormal dreams2189Risk Ratio (M-H, Fixed, 95% CI)2.91 [1.38, 6.14]
14 The number of patients experiencing abnormal dreams2189Odds Ratio (M-H, Fixed, 95% CI)3.55 [1.50, 8.37]
Analysis 1.1.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 1 The mean change of total recall on the Selective Reminding Test (SRT) from baseline.

Analysis 1.2.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 2 The mean total recall on the Selective Reminding Test (SRT) at exit.

Analysis 1.3.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 3 The mean change of total correct score on the 10/36 Spatial Recall Test (10/36 SRT) from baseline.

Analysis 1.4.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 4 The mean total correct score on the 10/36 Spatial Recall Test (10/36 SRT) at exit.

Analysis 1.5.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 5 The mean change of total correct score on the Symbol Digit Modalities Test (SDMT) from baseline.

Analysis 1.6.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 6 The mean total correct score on the Symbol Digit Modalities Test (SDMT) at exit.

Analysis 1.7.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 7 The mean change of total correct score on the Paced Auditory Serial Addition Test (PASAT) (2+3 sec) from baseline.

Analysis 1.8.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 8 The mean total correct score on the Paced Auditory Serial Addition Test (PASAT) (2+3 sec) at exit.

Analysis 1.9.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 9 The number of patients experiencing diarrhoea.

Analysis 1.10.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 10 The number of patients experiencing diarrhoea.

Analysis 1.11.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 11 The number of patients experiencing nausea.

Analysis 1.12.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 12 The number of patients experiencing nausea.

Analysis 1.13.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 13 The number of patients experiencing abnormal dreams.

Analysis 1.14.

Comparison 1 Subgroup analysis for donepezil versus placebo, Outcome 14 The number of patients experiencing abnormal dreams.

Appendices

Appendix 1. Keywords for searching the Cochrane MS Group Specialised Register

{memory disorder} OR {memory loss\*} OR {memory deficit} OR {memory impairment\*} OR {memory decline} OR {memory} OR {dysmnesia} OR {amnesia} OR {cognitive deficit\*} OR {cognitive disorder\*} OR {cognitive impairment\*} OR {cognitive dysfunction\*} OR {cognitive} OR {cognition} OR {cognitive decline} OR {dementia}

Appendix 2. PsycINFO (1980 to 26 June 2013)

#1 Multiple Sclerosis.sh.

#2 exp Myelitis.sh.

#3 Demyelinating Diseases*.tw.

#4 Encephalomyelitis.sh.

#5 Optic Neuritis.tw.

#6 Neuromyelitis Optica.tw.

#7 Transverse myelitis.tw.

#8 Devic.tw.

#9 Demyelinating disease*.tw.

#10 Acute disseminated encephalomyelitis.tw.

#11 1 OR 2 OR 3 OR 4 OR 5 OR 6 OR 7 OR 8 OR 9 OR 10 OR 11

#12 exp Memory Disorders

#13 memor*.mp.

#14 Dymnesi*.tw.  

#15 exp Amnesia.sh.

#16 (cogniti* adj3 (disorder* OR dysfunction* OR declin* OR deficit*)).tw.

#17 exp Dementia.sh.

#18 13 OR 14 OR 15 OR 16 OR 17 OR 18 OR 19 OR 20 OR 21 OR 22 OR 23 OR 24

#19 11 AND 18

#20 (198* OR 199* OR 200* OR 201*).yr, up.

#21 treatment effectiveness evaluation.sh.

#22 clinical trials.sh.

#23 mental health program evaluation.sh.

#24 placebo.sh.

#25 placebo$.ti,ab.

#26 randomly.ab.

#27 randomi#ed.ti,ab.

#28 trial.ti,ab.

#29 ((singl$ or doubl$ or trebl$ or tripl$) adj3 (blind$ or mask$ or dummy)).mp.

#30 (control$ adj3 (trial$ or study or studies or group$)).ti,ab.

#31 factorial$.ti,ab.

#32 allocat$.ti,ab.

#33 assign$.ti,ab.

#34 volunteer$.ti,ab.

#35 (crossover$ or cross over$).ti,ab.

#36 (quasi adj (experimental OR random$)).mp.

#37 "2000".md.

#38 22 OR 23 OR 24 OR 25 OR 26 OR 27 OR 28 OR 29 OR 30 OR 31 OR 32 OR 33 OR 34 OR 35 OR 36 OR 37 OR 38

#39 19 AND 20 AND 38

Appendix 3. CBMdisc (1978 - 25 June 2013)

#1 多发性硬化(MeSH)

#2 多发性硬化(TI/TW/AB/CT)

#3 视神经脊髓炎 (MeSH)

#4 视神经脊髓炎(TI/TW/AB/CT)

#5 脱髓鞘疾病(MeSH)

#6 脱髓鞘疾病(TI/TW/AB/CT)

#7 认知障碍(MeSH)

#8 认知(MeSH)

#9 认知?(TI/TW/AB/CT)

#10 记忆障碍(MeSH)

#11 记忆(MeSH)

#12 记忆?(TI/TW/AB/CT)

#13 遗忘(MeSH)

#14 遗忘(TI/TW/AB/CT)

#15 痴呆(MeSH)

#16 痴呆(TI/TW/AB/CT)

#17 #1 OR #2 OR #3 OR #4 OR #5 OR #6

#18 #7 OR #8 OR #9 OR #10 OR #11 OR #12 OR #13 OR #14 OR #15 OR #16

#19 #17 AND #18

What's new

Last assessed as up-to-date: 25 July 2013.

DateEventDescription
24 July 2013New search has been performedUpdated search run 24 July 2013.
24 July 2013AmendedThe review author team was reformed.
24 July 2013New citation required but conclusions have not changedThree new trials have been included in this review.

Contributions of authors

All correspondence: Dian He and Hongyu Zhou.
Drafting of review versions: Dian He, Shuai Dong.
Search for trials: Dian He and Yun Zhang.
Obtaining copies of trial reports: Dongfeng Wang and Xiangdong Gao.
Selection of trials for inclusion/exclusion: Dian He, Shuai Dong.
Extraction of data: Dian He, Shuai Dong.
Entry of data: Dian He, Hongyu Zhou.
Interpretation of data analyses: Dian He, Hongyu Zhou.

Declarations of interest

None known.

Differences between protocol and review

In the current review, types of participants were expanded by adding the new McDonald's criteria (Polman 2011). We excluded quasi-randomised trials in types of studies to avoid potential biases on results. We replaced the word "pharmacologic" with "pharmacological" in title and rephrased the primary outcome on efficacy (memory performance at week 12 (or later)) because they are more reasonable. The time of assessment (at week 12 (or later)) was specified to each outcome. All these changes from protocol to review have no effect on the review's conclusions.

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Krupp 2004

Methods

Randomised, double-blind, placebo-controlled clinical trial with parallel design in a single-centre.

The sample size and power were estimated.

35 patients were randomised to donepezil and 34 to placebo.

Donepezil group: 2 patients discontinued study drug due to AEs. 1 patient did not return at study end and was lost to follow-up.

Placebo group: 3 patients discontinued study drug due to AEs. 1 patient did not return at study end and was lost to follow-up.

34 patients in donepezil group and 33 patients in placebo group completed the 24-week follow-up visits.

35 patients in donepezil group and 34 in placebo group were analysed.

Treatment duration: 24 weeks.

Follow-up period: 24 weeks.

ITT analysis: yes.

Participants

Inclusion criteria: (1) the diagnosis of definite MS according to Poser's diagnostic criterion; (2) aged 18-55 years; (3) had to exhibit stable neurological functioning for at least 30 days prior to study entry and agree to continue their current medications for the study duration; (4) display at least mild verbal memory impairment, operationally defined as a score at least 0.5 SD below age- and gender-based normative data on the RAVLT; (5) had an absence of severe cognitive impairment on the MMSE (≥ 26); (6) all participants were ambulatory and had EDSS scores of ≤ 6.5; (7) concurrent antidepressants, antispasticity agents and DMTs (interferon-β or glatiramer acetate) were permitted, so long as dosage had been constant for at least 1 month prior to the evaluation.

Exclusion criteria: (1) evidenced severe depressive symptoms on the MADRS (≥ 14); (2) currently taking benzodiazepines were excluded, as these medications may affect cognition; (3) current alcohol or substance abuse, history of head injury or other medical condition known to affect cognition.

260 individuals were initially screened, 81 of whom were confirmed eligible. Potential participants were most commonly excluded because they refused entry (n = 72: scheduling difficulties, transportation issues, or lack of interest) or scored too highly on the RAVLT learning and memory screen (n = 59). Very few of the participants screened were excluded due to severe cognitive impairment, as indicated by low MMSE scores (n = 4). 35 patients were randomised to donepezil and 34 to placebo.

Most of characteristics of patients at baseline were similar except for gender, mean EDSS and MS subtypes.

Summary of patient characteristics at baseline (placebo: G1 (n = 34), donepezil: G2 (n = 35)):

  • Mean age: G1 = 45.85 ± 7.65 years, G2 = 42.49 ± 9.27 years

  • Female: G1 = 27 (79%), G2 = 20 (57%)

  • Mean education: G1 = 14.82 ± 2.32 years, G2 = 14.40 ± 2.37 years

  • Full-time employed: G1 = 14 (41%), G2 = 19 (54%)

  • Mean EDSS: G1 = 4.25 ± 1.62, G2 = 3.14 ± 1.78

  • MS subtypes: RRMS: G1 = 14 (41%), G2 = 24 (69%); SPMS: G1 = 19 (56%), G2 = 9 (26%); PPMS: G1 = 1 (3%), G2 = 2 (6%)

  • Years since symptom onset: G1 = 10.32 ± 6.78, G2 = 9.89 ± 8.50

  • Years since diagnosis: G1 = 9.12 ± 6.89, G2 = 7.09 ± 6.89

  • Concomitant medications: interferon-β: G1 = 23 (68%), G2 = 28 (80%); antidepressant medication: G1 = 13 (38%), G2 = 15 (43%); bladder medication: G1 = 10 (29%), G2 = 6 (17%); antispasticity medication: G1 = 13 (38%), G2 = 8 (23%); anticonvulsant medication: G1 = 10 (29%), G2 = 7 (20%)

  • Mean MADRS: G1 = 6.15 ± 3.62, G2 = 6.51 ± 3.31

  • Mean PANAS positive affect: G1 = 28.43 ± 6.05, G2 = 28.12 ± 8.46

  • Mean PANAS negative affect: G1 = 14.35 ± 5.19, G2 = 13.81 ± 5.10

  • Mean FSS: G1 = 5.06 ± 1.18, G2 = 4.97 ± 1.49

  • Mean MMSE: G1 = 28.44 ± 1.26, G2 = 28.71 ± 1.13

  • Mean WRAT3 Reading: G1 = 48.47 ± 3.29, G2 = 47.26 ± 4.10

  • Mean baseline SRT: G1 = 42.74 ± 8.78, G2 = 42.3 ± 9.03.

Interventions

Experimental intervention: the initial dose of donepezil was 5 mg/day, increasing to 10 mg/day at week 4 (G2)

Control intervention: placebo (G1)

Outcomes

Primary outcome: change score in total recall on the SRT (week 24 to week 0).

Secondary outcomes: (1) the patient self reported impression of memory change and the physician global impression of cognitive change, with a single physician completing all clinical assessments; (2) self reported changes in affect (PANAS) and fatigue (FSS); (3) BRB in MS: the 10/36 SRT total correct, SDMT (oral version) total correct, PASAT total correct sum of the 2 and 3 second forms of the task, and modified version of COWAT (Category Fluency) using mean number of correct words in the categories of animals and fruits/vegetables; (4) mean TOH performance; (5) AEs.

Notes

The study was registered with Clinicaltrials.gov NCT00062972 and approved by Stony Brook University's Committee on Research Involving Human Subjects. Patients were recruited from September 1999 to May 2002, primarily from Suffolk County, NY.

The evaluating physician was the same as the treating physician.

Groups differed at baseline on gender, MS type and EDSS.

Consistent with the selection of people having difficulty on the RAVLT verbal learning/memory test, participants performed most poorly at baseline on the verbal learning/memory measures of the SRT, performing 1.3-2.0 SD below norms, generally consistent with the classification of mild cognitive impairment.

A higher percentage of donepezil patients correctly guessed their medication status, as compared with placebo patients (rate ratio 1.77, 95% CI 1.00 to 3.11; P value = 0.038).

Groups did not differ in relapse rate during the study (P value = 0.329).

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "The randomization plan was generated by the study biostatistician using a computerized random number generator. Eligible patients were assigned to groups according to a simple randomization scheme using two blocking constraints. The randomization scheme was modified to a stratified permuted block design by sex randomization."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "The pharmacist was informed of all randomization assignments and was responsible for labelling the study drug and maintaining a master list linking the patients and their treatment assignments. The masking of the active and placebo treatments was preserved by creating treatments that looked identical."

Comment: allocation concealment was adequate.

Blinding (performance bias and detection bias)
All outcomes
Unclear risk

Quote: "All clinical staff and patients were masked regarding treatment assignment. One individual acted as both the evaluating physician and treating physician."

Comment: participants and personnel in this study were blinded to the allocated interventions. However, the evaluating physician was the same person as the treating physician who might have learned about any adverse events that the participant might have experienced due to study medication. This may have resulted in the possibility of detection bias.

Incomplete outcome data (attrition bias)
All outcomes
Low risk

According to the Subject Flow Figure and the corresponding description, 61 patients were randomly assigned to donepezil and 59 to placebo. 9 patients in donepezil group and 6 in placebo group discontinued study medication, among them, 3 patients in donepezil group and 4 in placebo group lost to follow-up. Finally, 58 in donepezil group and 55 in placebo group completed their final visit and data collection.

Comment: the reasons for drop-out were not carefully recorded; however, the rate of drop-outs (14.8% and 10.2% in donepezil group and placebo group, respectively) and the amounts of missing data were low. The difference in proportions between groups was small. Moreover, a last-observation-carried forward imputation strategy was used for missing data and the ITT principle was used for analyses. The risk for attrition bias was low.

Selective reporting (reporting bias)Low riskAll listed outcomes were reported adequately.
Other biasUnclear risk

Quote: "A higher percentage of donepezil subjects correctly guessed their medication status, as compared to placebo subjects (rate ratio, 1.77; 95% CI, 1.00 to 3.11; P=0.038)."

Comment: this may have led to biased results.

Quote: "Two treatment groups differed at baseline on sex, MS type and EDSS."

Comment: this may have led to biased results.

Krupp 2011

Methods

Multicentre, randomised, double-blind, placebo-controlled clinical trial with parallel design.

This study was powered on the basis of the findings for the primary outcome in Krupp 2004 (change in the SRT sum of recall) (sample sizes of 34 for placebo and 35 for donepezil). With actual sample sizes of 61 for donepezil and 59 for placebo, this study had a power of 79.1%.

61 patients were randomly assigned to donepezil and 59 to placebo.

Donepezil group: 9 (14.8%) patients discontinued study medication, among them, 3 patients lost to follow-up. 58 (95.1%) patients completed the study.

Placebo group: 6 (10.2%) patients discontinued study medication, among them, 4 patients lost to follow-up. 55 (93.2%) patients completed the study.

61 patients in donepezil group and 59 in placebo group were analysed.

Treatment duration: 24 weeks.

Follow-up period: 24 weeks.

ITT analysis: yes.

A last-observation-carried-forward imputation strategy was used for missing data.

Participants

Inclusion criteria: (1) had a clinically definite MS diagnosis according to McDonald's diagnostic criterion (2001 version); (2) aged 18-59 years; (3) had EDSS score ≤ 7.0; (4) participants could not have received steroids within 4 weeks of screening; (5) all MS subtypes were eligible; (6) participants had to score 0.5 SD below age- and gender-corrected normative data on the RAVLT; (7) concurrent use of antidepressants, antispasticity agents and DMTs were permitted; (8) participants had to agree to maintain stable doses of all medications to the extent possible.

Exclusion criteria: (1) benzodiazepine use was a basis for exclusion as these medications can affect cognition; (2) prior use of donepezil; (3) a current diagnosis of major depression; (4) current alcohol or substance abuse; (5) history of any other neurological or medical condition that could adversely affect cognition.

247 people with MS were screened for the study and 120 were enrolled. The most common reasons for not meeting eligibility criteria were (1) RAVLT above cut-off (54%); (2) declined participation (13%); (3) exclusionary medical issue (11%); (4) depression (10%); (5) benzodiazepine use (6%); (6) other (6%). 59 patients were assigned placebo treatment and 61 were assigned donepezil treatment. Of the 120 enrolled participants, 113 completed their final visit and data collection.

Groups did not differ significantly on most demographic or baseline features. However, the donepezil group had more years of education (P value = 0.027), higher reading scores on the WRAT 3 (P value = 0.026), and a longer self reported disease duration as measured from symptom onset (P value = 0.039). Self report on the OFQ indicated that a total of 38% of those on placebo and 44% of those on donepezil were on disability. OFQ responses also indicated that cognitive challenges associated with MS contributed to disability in 76% of the placebo and 65% of the donepezil group. No baseline differences between the donepezil and placebo groups reached significance on any of the neuropsychological test measures or questionnaires.

Summary of patient characteristics at baseline (placebo: G1 (n = 59), donepezil: G2 (n = 61)):

  • Demographics: mean age: G1 = 47.3 ± 8.9 years, G2 = 46.2 ± 7.5 years; women: G1 = 48 (81%), G2 = 45 (74%); mean education: G1 = 13.2 ± 2.0 years, G2 = 14.0 ± 2.2 years

  • MS characteristics: mean EDSS score: G1 = 3.74 ± 1.98%, G2 = 3.96 ± 1.78%; RRMS: G1 = 37 (63%), G2 = 38 (62%); SPMS: G1 = 19 (32%), G2 = 18 (30%); PPMS (%): G1 = 3 (5%), G2 = 5 (8%); years since symptom onset: G1 = 11.8 ± 8.0, G2 = 14.9 ± 8.2; years since diagnosis: G1 = 9.4 ± 7.6, G2 = 11.3 ± 7.7

  • Concomitant medications: interferon-β: G1 = 24 (41%), G2 = 28 (46%); glatiramer acetate: G1 = 9 (15%), G2 = 18 (30%); anticholinergic medications: G1 = 10 (17%), G2 = 14 (23%); non-tricyclic antidepressants: G1 = 26 (44%), G2 = 23 (38%); anticonvulsant medication: G1 = 14 (24%), G2 = 12 (20%)

  • Mean reading score on WRAT 3: G1 = 45.0 ± 6.5, G2 = 47.1 ± 3.5

  • Mean patient total on MSNSQ (SD): G1 = 30.3 ± 10.5, G2 = 30.2 ± 10.8

  • Mean clinician total on MSNSQ (SD): G1 = 25.5 ± 10.3, G2 = 25.7 ± 9.9

  • Mean scores on SRT (SD): G1 = 36.0 ± 9.8, G2 = 38.0 ± 8.7

  • Median scores on SRT (range): G1 = 36.0 (16-60), G2 = 40.0 (19-57)

  • Mean scores on RAVLT (SD): G1 = 37.5 ± 7.3, G2 = 35.6 ± 6.7

  • Median scores on RAVLT (range): G1 = 39.0 (17-49), G2 = 36.0 (18-47)

  • Primary occupational role: paid worker: G1 = 21 (36%), G2 = 15 (25%); disabled: G1 = 26 (44%), G2 = 23 (38%); cognitive challenges contribute to disability: G1 = 20 (76% of disabled), G2 = 15 (65% of disabled)

  • Psychological characteristics: mean mood sum on CMDI: G1 = 17.0 ± 7.7, G2 = 17.5 ± 8.0; mean mood T score on CMDI: G1 = 54.2 ± 14.0, G2 = 55.2 ± 14.5; mean total on FSS: G1 = 4.9 ± 1.5, G2 = 5.3 ± 1.4; mean total on Apathy Evaluation Scale Self Report: G1 = 31.7 ± 7.7, G2 = 31.5 ± 8.6.

Interventions

Experimental intervention: the initial dose was donepezil 5 mg/day, increased to 10 mg/day at week 4 (G2).

Control intervention: placebo (G1)

Outcomes

Primary outcome: (1) change in total recall on the SRT; (2) self reported memory change.

Secondary outcomes: (1) the10/36 Spatial Recall Test; (2) the SDMT total; (3) the PASAT 2 and 3 form total; (4) the COWAT letter fluency total; (5) Judgment of Line Orientation; (6) D-KEFS Sorting Test; (7) participant's impression of overall cognitive (not just memory) change; (8) impressions of memory and cognitive functioning from the perspective of the participant's partner (when available) and the evaluating clinician; (9) the CMDI; (10) the OFQ.

Notes

The clinical trial was registered with clinicaltrials.gov (protocol ID:Teva186557-1).

The mean impairment of enrolled participants was 1.7 ± 0.9 SD below norms on the RAVLT, similar to a prior donepezil study in MS. 42.5% of enrolled participants performed below fifth percentile of published norms (≥ 1.65 SD below the mean).

Missing data and reasons were not carefully recorded.

30% of placebo recipients and 40% of donepezil recipients treated correctly guessed their group assignment.

RCT had a multicentre design and a larger sample size than the earlier positive trial, though it had the same eligibility criteria and primary outcome.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "Using a random number generator, participants were assigned by the pharmacist at the Stony Brook University Hospital site to receive either donepezil or placebo in a 1:1 ratio, using a randomization scheme stratified by gender, MS type, and study site."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "The pharmacists at each site were the only individuals informed of the randomization assignments and were responsible for labelling study drug and maintaining a master list linking patients with treatment assignments."

Comment: allocation concealment was adequate.

Blinding (performance bias and detection bias)
All outcomes
Low risk

Quote: "Masking of active and placebo treatments was preserved by creating capsules that appeared identical. All other research staff and participants were masked regarding treatment assignment. The treating neurologist assessed each patient during the study visits. A separate evaluating clinician performed a structured interview to assess cognition at baseline and at the end-of-study visit. Medication compliance and adverse events were monitored by telephone at weeks 8, 12, 16, and 20, and in-person visits with the treating neurologist at weeks 4 and 14. The evaluating clinician was allowed only to ask about cognitive functioning and was specifically masked to any adverse events that the participant might have experienced due to study medication."

Comment: participants, personnel and outcomes assessors in this study were blinded to the allocated interventions.

Incomplete outcome data (attrition bias)
All outcomes
Low risk

According to the Subject Flow Figure and the corresponding description, 61 patients were randomly assigned to donepezil and 59 to placebo. 9 patients in donepezil group and 6 in placebo group discontinued study medication, among them, 3 patients in donepezil group and 4 in placebo group lost to follow-up. Finally, 58 in donepezil group and 55 in placebo group completed their final visit and data collection.

Comment: the reasons for drop-out were not carefully recorded; however, the rate of drop-outs (14.8% and 10.2% in donepezil group and placebo group, respectively) and the amounts of missing data were low. The difference in proportions between groups was small. Moreover, a last-observation-carried forward imputation strategy was used for missing data and the ITT principle was used for analyses. The risk for attrition bias was low.

Selective reporting (reporting bias)Low riskAll listed outcomes were reported adequately.
Other biasUnclear risk

Quote: "The donepezil group had more years of education (P= 0.027), higher reading scores on the Wide Range Achievement Test 3 (WRAT 3) (P= 0.026), and a longer self-reported disease duration as measured from symptom onset (P=0.039)."

Comment: this may have led to biased results.

Quote: "A total of 30% of placebo recipients and 40% of donepezil recipients treated correctly guessed their group assignment."

Comment: this may have led to biased results.

Lovera 2007

Methods

Randomised, double-blind, placebo-controlled trial with parallel design.

A formal sample size calculation or power analysis was not done.

21 patients were randomised to GB and 22 to placebo.

GB group: 1 person lost to follow-up for personal reasons.

Placebo group: 2 people lost to follow-up (1 due to pregnancy, 1 because of a traumatic injury and 1 because of a gastrointestinal illness).

20 people in the GB group and 19 in the placebo group completed the study. Available data for 20 people in the GB group (except for UFOV analysis, 1 patient was excluded because of unreliable observation) and 19 in the placebo group (except for Stroop analysis, 1 patient was excluded because they were colour blind) were analysed.

Treatment duration: 12 weeks.

Follow-up period: 17 weeks. (The data from the visit at week 5 served as the baseline. Treatment with GB or placebo was started after the week 5 visit.)

ITT analysis: no.

Participants

Inclusion criteria: (1) diagnosis of MS by McDonald's criteria; (2) aged 18-60 years; (3) a score 0.5-2.5 SD below the mean on the PASAT or the CVLT-II total score.

Exclusion criteria: (1) history of alcoholism, drug abuse, psychosis, major depression or a score on the BDI-IA > 19; (2) people with severe cognitive impairment who required a permanent carer; (3) any significant medical problem, including insulin-dependent diabetes, uncontrolled hypertension, uncontrolled hypothyroidism, hepatic disease, renal disease, pulmonary disease, congestive heart failure, angina, valvular disease or abnormalities of coagulation; (4) a significant MS exacerbation within the 30 days before screening; (5) current use of anticoagulants; (6) pregnancy; (7) any visual problem that would interfere with testing, such as corrected binocular visual acuity worse than 20/50, nystagmus on primary gaze or impaired colour vision; (8) not speaking English as the native language; (9) any other condition that the investigator thought would make them unsuitable for the study.

68 people were screened and 43 met inclusion/exclusion criteria. Of the 25 who did not meet entry criteria, 19 scored too well on the PASAT and CVLT-II, 4 had BDI scores > 19 and 2 had medical illnesses that excluded them from the trial. Of the 43 who met the inclusion criteria, 21 were randomised to GB and 22 to placebo.

There were no significant differences on the demographic characteristics and the use of psychoactive or sedating medications between groups. There were no differences in performance on any of the tests of the neuropsychological test battery at baseline.

Summary of patient characteristics at baseline: (placebo: G1 (n = 19), GB: G2 (n = 20)):

  • Gender: male: G1 = 6 (32%), G2 = 6 (30%); female: G1 = 13 (68%), G2 = 14 (70%)

  • Race: white: G1 = 19 (100%), G2 = 18 (90%); other: G1 = 0 (0), G2 = 2 (10%)

  • Type of MS: RRMS: G1 = 11 (58%), G2 = 11 (55%); PPMS: G1 = 2 (10%), G2 = 0 (0); SPMS: G1 = 6 (32%), G2 = 9 (45%)

  • Education: high school: G1 = 2 (11%), G2 = 5 (25%); college: G1 = 12 (63%), G2 = 11 (55%); master's degree or higher: G1 = 5 (26%), G2 = 4 (20%)

  • Age: G1 = 50.2 ± 8.0 years, G2 = 47.8 ± 6.7 years

  • Years since MS onset: G1 = 15.6 ± 6.3, G2 = 15.1 ± 8.8

  • EDSS score: G1 = 3.5 ± 1.3, G2 = 3.8 ± 1.6

  • 9-HPT: G1 = 27.2 ± 10.2 s, G2 = 26.4 ± 11.2 s

  • 25 Foot Timed Walk: G1 = 5.8 ± 1.2 s, G2 = 6.8 ± 3.7 s

  • BDI-IA: G1 = 6.9 ± 5.4, G2 = 7.7 ± 5.1.

Number of people using different psychoactive or sedating medications in each group (n):

  • Selective serotonin re-uptake inhibitors: G1 = 6, G2 = 6

  • Tricyclic antidepressants: G1 = 3, G2 = 3

  • Benzodiazepines: G1 = 3, G2 = 3

  • Stimulants: G1 = 6, G2 = 6

  • Anticonvulsants: G1 = 3, G2 = 3

  • Narcotics: G1 = 1, G2 = 1

  • Baclofen: G1 = 4, G2 = 5

  • Anticholinergics: G1 = 4, G2 = 5.

Interventions

Experimental intervention: GB 120 mg twice a day (G2).

Control intervention: placebo (G1).

Outcomes

Primary outcomes: (1) free delayed recall score from the CVLT-II; (2) time for the colour-word interference condition from the Stroop; (3) threshold from an adapted version of the UFOV, (4) scores on the SDMT; (5) scores on the COWAT; (6) scores on the 3 second version of the PASAT.

Secondary outcomes: (1) the other subscores from the tests in the cognitive battery; (2) perceived cognitive deficits assessed by the PDQ; (3) quality of life assessed by the MSQLI); (4) safety (the frequency of side effects).

Notes

This is the first study that evaluated the effects of GB on cognitive performance. The Oregon Health and Science University (OHSU) Institutional Review Board approved the study protocol and informed consent. Recruitment started in July 2002 and ended in June 2004.

41% of people taking GB and 59% taking placebo guessed their treatment allocation correctly.

No information on DMTs was reported.

The GB (120 mg) and placebo capsules were provided by Thorne Research Inc. (Dover, ID, USA).

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "The allocation sequence was generated using a random number generator. The sequence was stratified by age and had a block size of six."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "The staff at the OHSU Medical Informatics and Clinical Epidemiology Department generated the allocation sequence, and numbered sequentially the containers provided by the manufacturer accordingly."

Comment: allocation concealment was adequate.

Blinding (performance bias and detection bias)
All outcomes
Low risk

Quote: "The blinding was implemented by using identical containers, instructions and labels. The subjects, physicians and research assistants did not know the treatment allocation at any point until the data analysis was completed. The rest of the study personnel did not have access to the allocation sequence until the study was completed. A neuropsychologist trained the research assistants in the administration of the battery. A research assistant obtained side-effect reports by telephone at four and eight weeks after starting treatment and at exit. One of the study physicians evaluated the side-effect reports and graded them."

Comment: participants, personnel and outcomes assessors in this study were blinded to the allocated interventions.

Incomplete outcome data (attrition bias)
All outcomes
High risk

According to the Subject Flow Figure and the corresponding description, 21 people were randomised to GB and 22 to placebo. 1 person in GB group lost to follow-up, and 3 people in placebo group lost to follow-up. 20 people in the GB group and 19 in the placebo group completed the study. Available data for 20 participants in the GB group (except for UFOV analysis, 1 patient was excluded because of unreliable observation) and 19 in the placebo group (except for Stroop analysis, 1 patient was excluded because they were colour blind) were analysed.

Comment: the rate of drop-outs (4.8% and 13.6% in the GB and placebo group, respectively) and the amounts of missing data were low, and the reasons for missing outcomes were carefully recorded and balanced between groups. However, the difference in proportions between groups was large; moreover, the ITT principle was not used for analyses. The risk for attrition bias was high.

Selective reporting (reporting bias)Unclear riskAll listed outcomes were reported, but some outcomes were lack of data.
Other biasUnclear risk

Quote: "A total of 41% of subjects taking GB and 59% of the subjects taking placebo guessed their treatment allocation correctly."

Comment: this may have led to biased results.

No information was reported on DMTs, this may have led to biased results.

Lovera 2010

Methods

Double-blind, randomised, placebo-controlled trial in 4 MS centres.

The sample size and power were estimated. With only 2 tests as the primary outcome measure, the subsequent univariate tests with Bonferonni correction had > 80% power.

58 patients were randomly allocated to memantine and 68 to placebo. 4 patients in memantine group did not receive allocated intervention due to (1) protocol violation and history of seizures; (2) decided not to participate; (3) time commitment; (4) not recorded. 3 patients in placebo group did not receive allocated intervention due to (1) not recorded; (2) decided not to participate; (3) did not attend for follow-up visits.

54 patients in memantine group and 65 in placebo group received allocated intervention.

Memantine group: 9 patients discontinued intervention due to (1) somnolence; (2) nausea, root canal treatment; (3) UTI, insomnia; (4) memory changes, insomnia, ataxia, dysarthria; (5) agitation, dizziness; (6) nausea, vomiting; (7) dizziness, MS relapse; (8) rash; (9) headache. 0 lost to follow-up.

Placebo group: 5 patients discontinued intervention due to (1) flatulence; (2) cold; (3) leg fracture; (4) UTI, increased memory problems; (5) unknown. 4 patients lost to follow-up due to (1) mood changes; (2) headaches; (3) headaches; (4) did not return calls.

54 patients were analysed in memantine group, and 60 patients were analysed in placebo group (1 patient had lost baseline PASAT data).

Treatment duration: 16 weeks.

Follow-up period: 16 weeks.

ITT analysis: no.

Participants

Inclusion criteria: (1) MS as per McDonald's criteria determined by 1 of the study physicians after review of the patient's medical record and imaging (participants with all types of MS were eligible); (2) aged 18-65 years; (3) subjective cognitive complaints and a score on the PASAT or the CVLT-II (LDFR or total recall) worse than 1 SD below the mean from appropriate age- and education-adjusted norms.

Exclusion criteria: (1) history of major depression or psychosis, or a score > 19 on the BDI-IA; (2) corrected binocular visual acuity worse than 20/50 or impairment of binocular colour vision; (3) a primary language different from English; (4) pregnancy; (5) renal insufficiency; (6) seizures; (7) use of medical marijuana or abuse of alcohol or other illegal drugs in the prior 6 months; (8) use of medications that could interact with memantine; (9) an MS relapse in the prior 30 days.

206 people were initially screened, 75 did not meet inclusion criteria due to (1) BDI; (2) PASAT and CVLT; (3) impaired vision; (4) unable to read. 4 refused to participate. 1 was not recorded. 58 patients were randomly allocated to memantine and 68 to placebo.

After allocation, 4 in memantine group did not receive allocated intervention due to (1) protocol violation, history of seizures; (2) decided not to participate; (3) time commitment; (4) not recorded. 3 in placebo group did not receive allocated intervention due to (1) not recorded; (2) decided not to participate; (3) did not attend for follow-up visits.

54 patients received memantine treatment and 65 received placebo treatment.

Groups were well matched on demographic characteristics except for type of MS. There was an imbalance on the z-scores for the CVLT-II LDFR, but well matched on the raw test scores.

Summary of patient characteristics at baseline (placebo: G1 (n = 68), memantine: G2 (n = 58)):

  • Age: G1 = 50.4 ± 7.7 years, G2 = 50.5 ± 8.2 years

  • Female: G1 = 50 (77%), G2 = 45 (83%)

  • Education: G1 = 14.6 ± 2.2 years, G2 = 14.5 ± 2.1 years

  • EDSS: G1 = 4.4 ± 1.9, G2 = 4.5 ± 2.2

  • Disease duration: G1 = 13.3 ± 8.3 years, G2 = 14.0 ± 9.0 years

  • PASAT, # correct: G1 = 33.1 ± 13.1, G2 = 32.8 ± 13.7

  • PASAT, z-score: G1 = -1.7 ± 1.3, G2 = -1.8 ± 1.4

  • CVLT-II free delayed recall, # words: G1 = 8.1 ± 3.6, G2 = 7.2 ± 3.4

  • CVLT-II free delayed recall, z-score : G1 = -1.1 ± 1.2, G2 = -1.5 ± 1.3

  • Race: white: G1 = 61 (94%), G2 = 52 (98%); African-American: G1 = 2 (3%), G2 = 0 (0); other: G1 = 2 (3%), G2 = 2 (2%); Hispanic: G1 = 2 (3%), G2 = 1 (2%)

  • Type of MS: RRMS: G1 = 47 (72%), G2 = 28 (52%); PPMS: G1 = 6 (9%), G2 = 12 (22%); SPMS: G1 = 12 (18%), G2 = 14 (26%).

Interventions

Experimental intervention: memantine 10 mg twice a day (4-week titration followed by 12 weeks on the highest tolerated dose) (G2).

Control intervention: placebo (G1).

Outcomes

Primary outcome: the change from baseline to exit on (1) PASAT; (2) CVLT-II LDFR.

Secondary outcomes: (1) COWAT; (2) the Stroop Test (Victoria Version); (3) SDMT; (4) D-KEFS Sorting Test, self reports measures on (5) cognitive deficits using the PDQ; (6) quality of life using the SF-36; (7) fatigue using the MFIS and FSS; (8) depression using the BDI-IA; reports from the participant's closest family member or carer using (9) the MSNSQ and (10) the MNPI; (11) adverse events.

Notes

People were recruited from 4 MS centres in the US (Oregon Health & Science University, Portland, OR; Evergreen Healthcare Medical Center, Kirkland, WA; University of Texas Southwestern Medical Center, Dallas, TX; and University of Southern California (USC), Los Angeles, CA). Recruitment started in August 2004 and ended in October 2007. The study was registered with Clinicaltrials.gov NCT00300716 and IRB's at each site approved the study.

No information on DMTs was reported.

There were no comparisons at baseline between groups for the scores on the BDI, FSS and MFIS.

Group sizes were slightly imbalanced, which can occur when using several strata and blocking in the allocation sequence. There was 1 protocol violation where 1 person was administered the alternate forms of the PASAT in the incorrect order but this participant's data were still used.

The initial protocol required 3 visits over 3 weeks prior to starting therapy.

The baseline data for the PASAT on 1 of the people in the placebo group were lost.

A violation occurred where 1 person with history of seizures was randomised but the study team found the mistake prior to starting him on the study drug.

Participants were wrong regarding their treatment assignment more frequently than expected by chance (61%; P value = 0.05 binomial test) because they were more likely to think they had received placebo.

Two MS exacerbations occurred during the study, 1 in each group. There was no significant difference between the groups on the change on the EDSS. The proportion of people experiencing progression defined as 1 point worsening for EDSS > 4.5 and 0.5 points for EDSS 4.5 was not significantly different between the groups.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "An independent biostatistician at OHSU (Cynthia Morris, PhD) used EXCEL's random number generator to create the assignment sequence with a block size of six; stratification by centre, age, and score on the CVLT-II LDFR; and equal probability of assignment to either arm."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "The independent statistician's staff would receive a fax with the subject's study ID, randomize the subject, and then send a fax to the pharmacy indicating the treatment allocation.

Comment: allocation concealment was adequate.

Blinding (performance bias and detection bias)
All outcomes
Low risk

Quote: "Only the independent statistician's staff and the research pharmacists had access to the allocation sequence until the analysis was completed. Evaluating research assistants and physicians administered the cognitive and behavioural questionnaires and disability scales while treating physicians and research assistants evaluated adverse events and managed dose titration. The subjects and the evaluating team were instructed not to discuss any health related matters to maintain the evaluating team's blinding. The questionnaires administered at the end of the study indicate that blinding of the subjects, evaluating physicians and evaluating research assistants was successful."

Comment: participants, personnel and outcomes assessors in this study were blinded to the allocated interventions.

Incomplete outcome data (attrition bias)
All outcomes
High risk

According to the CONSORT flow chart, 54 patients received memantine, and 65 patients received placebo. 9 patients in memantine group discontinued intervention and 5 patients in placebo group discontinued intervention. 0 patients in memantine group and 4 in placebo group lost to follow-up. 45 patients in memantine group and 56 in placebo group completed the study. 54 patients in memantine group were analysed, and 60 were analysed in placebo group because 1 person lost baseline PASAT data).

Comment: the rate of drop-outs (22.4% and 17.6% in the memantine and placebo group, respectively) and the amounts of missing data were high. Moreover, the reasons for drop-out were carefully recorded but differed between groups. The study authors excluded the patients who were lost to follow-up and did not use the ITT principle for analyses. The risk for attrition bias was high. Generally, the potential impact of missing continuous outcomes increases with the proportion of participants with missing data.

Selective reporting (reporting bias)Unclear riskAll listed outcomes were reported, but some outcomes lacked data.
Other biasUnclear risk

Quote: "The groups were well matched on demographic characteristics except for type of MS. Although the groups were well matched on the raw test scores, there was an imbalance on the z-scores for the CVLT-II LDFR."

Comment: this may have led to biased results.

Quote: "Subjects were wrong regarding their treatment assignment more frequently than expected by chance (61%; P=0.05 binomial test) because they were more likely to think they had received placebo."

Comment: this may have led to biased results.

Lovera 2012

Methods

Randomised, double-blind, placebo-controlled clinical trial with parallel design.

The sample size and power were estimated. The researchers estimated needing 110 people with complete observations to have 80% to detect a difference as low as 0.5 on the z-scores on any of the tests at both the multivariate level and the univariate levels at an α level of 0.05 (2-sided). However, the researchers stopped enrolment after 121 participants had been enrolled because the dropout rate was less than 5% and more than 110 participants had completed the study.

61 patients were randomly allocated to GB and 60 to placebo. 1 patient in placebo group did not receive allocated intervention because of a randomisation error, the patient did not meet inclusion criteria as he refused to complete the PASAT.

61 patients in GB group and 59 in placebo group received allocated intervention.

GB group: 3 patients lost to follow-up (1 did not return telephone calls, 2 discontinued intervention due to AE (myocardial infarction and psychiatric hospitalisation) and did not complete exit visit). 1 discontinued intervention due to AE (nausea and heartburn) but completed exit visit.

Placebo group: 1 patient lost to follow-up (developed pruritus and discontinued intervention and did not complete exit visit).

61 patients were analysed in GB group, and 59 of patients were analysed in placebo group because 1 person lost baseline PASAT data.

Treatment duration: 12 weeks.

Follow-up period: 12 weeks.

Modified ITT analysis: yes.

Participants

Inclusion criteria: people (1) with MS by McDonald's criteria (2001 version); (2) had EDSS score 0-7.5; (3) were stable for at least 30 days; (4) scored ≥ 1 SD below the mean of established normative samples on ≥ 1 of the 4 tests in the neuropsychological test battery (Stroop (Victoria Version); COWAT; long-delay free-recall portion of the CVLT-II; and the 2 second version of the PASAT); (5) did not have contraindications for GB; (6) did not have problems that would interfere with testing.

173 individuals with MS were screened for the study and 121 were enrolled. The reasons for not meeting eligibility criteria were (1) cognitive scores too high (1 on second screening here) (n = 35); (2) BDI-Ⅱ scores too high (n = 7); (3) failed colour vision (n = 4); (4) had medications that were contraindicated (n = 2); (5) other (n = 3); (6) decided not to participate after qualifying (n = 1). Of the 121 enrolled participants, 61 patients were randomly allocated to GB and 60 to placebo.

Groups did not differ significantly on most demographic or baseline features except for gender.

Summary of patient characteristics at baseline (placebo: G1 (n = 59), GB: G2 (n = 61)):

  • Age: mean (SD) G1 = 53 ± 9.5 years, G2 = 51.3 ± 8.6 years; median (range) G1 = 56 (24-65) years, G2 = 53 (27-65) years

  • Female: G1 = 37 (63%), G2 = 29 (48%)

  • Education: did not complete high school: G1 = 1 (2), G2 = 0 (0); high school diploma or General Educational Development: G1 = 2 (3%), G2 = 4 (7%); some college: G1 = 26 (44%), G2 = 33(54%); college graduate: G1 = 17 (29%), G2 = 10 (16%); some graduate school: G1 = 3 (5%), G2 = 5 (8%); master's degree or higher: G1 = 10 (17%), G2 = 9 (15%)

  • Race: white: G1 = 59 (100%), G2 = 55 (90%); non-white: G1 = 0(0), G2 = 6 (10%)

  • Ethnicity: Hispanic or Latino: G1 = 2 (3%), G2 = 2 (3%)

  • EDSS score: mean (SD) G1 = 4 ± 2, G2 = 4 ± 2; median (range) G1 = 4 (1-7.5), G2 = 4 (0-7.5)

  • Disease duration: mean (SD) G1 = 19.3 ± 11.8 years, G2 = 20.9 ± 11.8 years; median (range) G1 = 18 (1-45) years, G2 = 20 (4-47) years

  • Type of MS: PPMS G1 = 4 (7%), G2 = 5 (8%); PRMS G1 = 1 (2%), G2 = 0 (0); RRMS G1 = 35 (59%), G2 = 42 (69%); SPMS G1 = 19 (32%), G2 = 14 (23%)

  • Centre: Seattle: G1 = 27 (46%), G2 = 29 (48%); Portland: G1 = 32 (54%), G2 = 32 (52%)

  • DMT: glatiramer acetate: G1 = 13 (22%), G2 = 20 (33%); no DMT: G1 = 16 (27%), G2 = 17 (28%); interferon β-1a 30 μg once a week: G1 = 9 (15%), G2 = 10 (16%); interferon β-1a 44 μg 3 times a week: G1 = 10 (17%), G2 = 5 (8%); natalizumab: G1 = 4 (7%), G2 = 3 (5%); interferon β-1b 0.25 mg every other day: G1 = 3 (5%), G2 = 2 (3%); mitoxantrone G1 = 1 (2%), G2 = 1 (2%); azathioprine G1 = 0 (0%), G2 = 1 (2%); interferon β-1a 30 μg/week changed to glatiramer acetate: G1 = 0 (0%), G2 = 1 (2%); mycophenolate: G1 = 1 (2%), G2 = 0 (0%); interferon β-1a 44 μg and methotrexate: G1 = 1 (2%), G2 = 0 (0%); interferon β-1a 44 μg and glatiramer acetate: G1 = 0 (0%), G2 = 1 (2%); interferon β-1a 44 μg and glatiramer acetate: G1 = 1 (2%), G2 = 0 (0%).

Interventions

Experimental intervention: GB 120 mg twice a day (G2).

Control intervention: placebo (G1).

Outcomes

Primary outcomes: z-scores based on healthy, age-matched control norms for (1) the time on Color-Word portion of the Stroop Test (Victoria Version); (2) COWAT; (3) LDFR portion of the CVLT-II; (4) the 2 second version of the PASAT.

Secondary outcomes: (1) perceived cognitive deficits assessed by the PDQ; (2) the MSNSQ; (3) CIQ; (4) fatigue assessed by the MFIS; (5) depression assessed by the BDI-II; (6) safety; (7) relapses.

Notes

The study was registered with clinicaltrials.gov (NCT00841321) and performed at 2 sites, the Portland and Seattle VA Medical Centers. The full protocol of the study can be accessed at www.ohsu.edu/ms/ginkgoprotocol (last accessed 10 December 2013).

GB: EGb-761, Willmar Schwabe GmbH & Co, Germany. The GB tablets contained 29.7 mg of flavoglycosides and 7.3 mg of terpene lactones.

The sample had mild impairment overall and mainly were impaired on the PASAT and mostly below average on the CVLT-II.

No comparisons at baseline between groups for the scores on the MFIS and the BDI-II.

Pill counts indicated that 88% of participants took 80% or more of their treatment.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "The Portland VA research pharmacist generated the allocation sequence in Excel with alternating random blocks of 4 or 6 subjects stratified by disease duration and study site and a 1:1 ratio."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "The Portland VA research pharmacist generated the allocation sequence in Excel with alternating random blocks of 4 or 6 subjects stratified by disease duration and study site and a 1:1 ratio. Only the research pharmacists had access to the allocation sequence."

Comment: allocation concealment was adequate.

Blinding (performance bias and detection bias)
All outcomes
Low risk

Quote: "The research pharmacists assigned the participants to either ginkgo or matching placebo tablets and dispensed the tablets to them in matching identical containers. One research assistant assessed adverse events and another administered the neuropsychological tests. They were instructed not to discuss adverse events between them to ensure that the cognitive test assessments would be masked."

Comment: participants, personnel and outcomes assessors in this study were blinded to the allocated interventions.

Incomplete outcome data (attrition bias)
All outcomes
Low risk

According to the CONSORT flow chart, 61 patients were randomly allocated to GB and 60 to placebo. 1 patient in placebo group did not receive allocated intervention because he refused to complete the PASAT. 3 patients in GB group and 1 in placebo group were lost to follow-up. 1 in GB group discontinued intervention due to AE. 61 patients were analysed in GB group, and 59 patients were analysed in placebo group because 1 patient lost baseline PASAT data.

Comment: the reasons for drop-out differed between groups; however, the rate of drop-outs (6.6% and 3.3% in GB group and placebo group, respectively) and the amounts of missing data were low, the difference in proportions between groups was small. Moreover, the modified ITT principle was used for analyses. The attrition bias was low.

Selective reporting (reporting bias)Low riskAll listed outcomes were reported adequately.
Other biasUnclear risk

Quote: "The patients were enrolled based on scoring 1 SD or more below the mean of established normative samples on one or more of the 4 tests in the neuropsychological test battery (the Stroop Color-Word Test; the COWAT; the CVLT-II; and the PASAT).The sample had mild impairment overall and mainly were impaired on the PASAT and mostly below average on the CVLT-II."

Comment: COWAT is used to measure language, therefore, selection of targeted sample is not specific for memory impairment.

Quote: "The randomization balanced the groups well except for gender."

Comment: this may have led to biased results.

Mäurer 2013

Methods

Randomised, placebo-controlled, multicentre trial used a double-blind, parallel groups design.

The sample size and power were estimated. Assuming a similar improvement in the placebo arm but a mean change of 4.59 with an SD of 8.7 in the active treatment group, a sample size of 90 in each group would have had 80% power to detect a difference in means of 3.02, using a 2-group Satterthwaite t test with a 0.05 2-sided significance level. Assuming a 10% drop-out, 100 patients needed to be randomised to each treatment arm. However, the study was prematurely terminated after the recruitment of 86 patients because of slow recruitment, with this recruitment the actual power of the study is 49%.

45 patients were randomised to rivastigmine and 41 to placebo. 18 of 86 patients enrolled (20.9%) prematurely discontinued the study:

Rivastigmine group: 11 (24.4%) patients prematurely discontinued the study due to AEs (n = 8), condition no longer required study drug (n = 1), withdrew consent (n = 1), lost to follow-up (n = 0), administrative problems (n = 1). 34 patients completed the double-blind study phase at week 16.

Placebo group: 7 (17.1%) patients prematurely discontinued the study due to adverse (n = 3), condition no longer required study drug (n = 0), withdrew consent (n = 2), lost to follow-up (n = 1), administrative problems (n = 1). 34 patients completed the double-blind study phase at week 16.

43 patients in rivastigmine group and 38 in placebo group were analysed.

Treatment duration: 68 weeks (including 4-week titration period and 12-week maintenance period (double-blind study phase), and 12-month open-label treatment phase).

Follow-up period: 68 weeks.

Modified ITT analysis: yes.

Participants

Inclusion criteria: individuals (1) had diagnosis of MS (subtypes RRMS, SPMS or clinically isolated syndrome, but not stratified a priori) as defined by the 2005 revised McDonald criteria; (2) aged 18-65 years; (3) had cognitive impairment as defined by a FST score ≥ 3.0, a MUSIC score of ≤ 19 at screening, or both (4) participants were eligible if they had received interferonβ-1b therapy for at least 60 days before baseline.

Exclusion criteria: patients (1) used medication for Alzheimer's disease; (2) started taking psychoactive medication; (3) used muscle relaxants or lithium at different time points before randomisation; (4) pregnant or breastfeeding; (5) with diabetes mellitus, malignancy, any cognition-affecting medical condition; (6) had drug addiction, alcohol abuse; (7) had depression as indicated by a MADRS score ≥ 14 at screening; (8) had been subjected to cognitive testing with BRB-N within the last year before randomisation; (9) had attended any cognitive rehabilitation study or program in the 3 months prior to the screening visit. During the study, centrally acting anticholinergics and medication for Parkinson's disease were not permitted.

86 of 125 MS patients screened were enrolled and randomised to either rivastigmine (n = 45) or placebo (n = 41). Of the 86 patients enrolled, 72 had RRMS, 13 SPMS and 1 had been diagnosed with clinically isolated syndrome. Of the 86 patients, 60 satisfied the criteria of cognitive impairment based on the results of the MUSIC and the FST test combined, 11 satisfied the criteria based only on the results of the MUSIC test, and 15 based only on the results of the FST test.

There were no statistically significant differences in baseline characteristics between the rivastigmine and placebo groups.

Summary of patient characteristics at baseline (placebo: G1 (n = 38), rivastigmine: G2 (n = 43)):

  • Mean age: G1 = 44.0 ± 7.3 years, G2 = 44.6 ± 9.4 years

  • Female: G1 = 20 (52.6%), G2 = 23 (53.5%)

  • Score on the MSFC: G1 = -0.03 ± 0.76, G2 = 0.11 ± 0.67

  • Score on the SRT total recall: G1 = 48.21 ± 10.32, G2 = 43.93 ± 10.67

  • Score on SRT long-term storage: G1 = 36.89 ± 13.69, G2 = 31.95 ± 14.41

  • Score on the MADRS: G1 = 8.21 ± 4.99, G2 = 6.63 ± 4.49

  • Score on the FST: G1 = 3.56 ± 1.14, G2 = 3.70 ± 1.17

  • Score on the MFIS: G1 = 37.89 ± 17.21, G2 = 35.74 ± 18.58

  • Score on Multiple Sclerosis Inventarium Cognition: G1 = 16.14 ± 5.29, G2 = 15.28 ± 5.29.

Interventions

Experimental intervention: rivastigmine patches of 5 cm2 (4.6 mg/day), increased to 10 cm2 (9.5 mg/day) at week 4 (G2).

Control intervention: placebo (G1).

Outcomes

Primary outcomes: (1) improvement in memory performance, as assessed by the total recall on the SRT; (2) safety.

Secondary outcomes: (1) improvement in activities of daily living, as assessed by a subscale of the PRIMuS and improvement in the score of the CGI scale; (2) additional BRB-N subtests (the SDMT; PASAT-3 seconds; 10/36 Spatial Recall Test); (3) changes in cognitive functioning, as assessed by subscales of the BRB, BRB-N composite index score, MFIS, MADRS, FST and MSFC.

Notes

The study started in January 2009, and participants were enrolled at 30 investigational sites in Germany. However, the study was terminated in January 2010 because of slow recruitment.

The study is registered in the clinicaltrials.gov database (protocol ID: NCT00881205) and supported by Novartis Pharma GmbH.

After a screening period of 6 weeks, participants were randomised to either rivastigmine or placebo in a 1:1 ratio. Participants then entered a 4-week titration period, followed by a 12-week maintenance period and an optional 12-month open-label treatment phase.

Results were presented for the modified ITT population. The modified ITT population consisted of all patients randomised who received at least 1 dose of study drug and had at least 1 baseline and post-baseline efficacy assessment. Safety was assessed in patients who received at least 1 dose of study drug and had at least 1 post-baseline safety assessment.

During the study, 4 of 35 (8.9%) rivastigmine-treated patients and 6 of 27 (14.6%) placebo-treated patients had an MS relapse.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "Randomisation lists were generated by an external clinical research organisation (CRO) using a validated system that automates the random assignment of the treatment arms under the responsibility of the GCP officer."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "Randomisation lists were generated by an external clinical research organisation (CRO) using a validated system that automates the random assignment of the treatment arms under the responsibility of the GCP officer. The investigator enrolled the subjects and assigned participants to treatment according to the randomisation list."

Comment: allocation concealment was adequate.

Blinding (performance bias and detection bias)
All outcomes
Low risk

Quote: "Patients, investigator staff, persons performing the assessments and data analysts remained blinded throughout the entire study period. Study drugs were identical in packaging, labelling, schedule of administration, appearance and odour."

Comment: participants, personnel and outcomes assessors in this study were blinded to the allocated interventions.

Incomplete outcome data (attrition bias)
All outcomes
High risk

According to the Subject Flow and the corresponding description, 45 patients were randomised to rivastigmine and 41 to placebo. 18 of 86 patients enrolled (20.9%) prematurely discontinued the study. 11 (24.4%) patients in rivastigmine group prematurely discontinued the study due to AEs, condition no longer required study drug, withdrew consent and administrative problems (n = 1). 34 patients completed the double-blind study phase at week 16. 7 (17.1%) patients in placebo group prematurely discontinued the study due to AEs, withdrew consent, lost to follow-up and administrative problems. 34 patients completed the double-blind study phase at week 16. 43 patients in rivastigmine group and 38 in placebo group were analysed.

Comment: although the modified ITT principle was used for analyses, the rate of drop-outs and the amounts of missing data were high and the reasons for drop-out differed between groups. Attrition bias was high. Generally, the potential impact of missing continuous outcomes increases with the proportion of participants with missing data.

Selective reporting (reporting bias)Low riskAll listed outcomes were reported adequately.
Other biasUnclear risk

The study was supported by Novartis Pharma GmbH.

Comment: this may have led to biased results.

The patients were enrolled based on cognitive impairment as defined by a FST score of ≥ 3.0, a MUSIC score of ≤ 19, or both at screening.

Comment: selection of targeted sample is not specific for memory impairment.

Shaygannejad 2008

  1. a

    9-HPT: 9 Hole Peg Test; AE: adverse effects; BDI: Beck Depression Inventory; BRB-N: Brief Repeatable Battery of Neuropsychological Tests; CGI: Clinical Global Impression; CI: confidence interval; CIQ: Community Integration Questionnaire; CMDI: Chicago Multiscale Depression Inventory; COWAT: Controlled Oral Word Association Test; CVLT-II: California Verbal Learning Test II; D-KEFS: Delis-Kaplan Executive Functions System; DMT: disease-modifying therapy; EDSS: Expanded Disability Status Scale; FSS: Fatigue Severity Scale; FST: Faces Symbol Test; G1: group 1; G2: group 2; GB: Ginkgo biloba; ITT: intention-to-treat; LDFR: Long Delay Free Recall; MADRS: Montgomery-Asberg Depression Scale; MFIS: Modified Fatigue Impact Scale; MMSE: Mini-Mental State Examination; MNPI: Modified Neuropsychiatric Inventory; MS: multiple sclerosis; MSFC: Multiple Sclerosis Functional Composite; MSNSQ: Multiple Sclerosis Neuropsychological Screening Questionnaire; MSQLI: Multiple Sclerosis Quality of Life Index; MUSIC: Multiple Sclerosis Inventarium Cognition; OFQ: Occupational Functioning Questionnaire; PANAS: Positive and Negative Affect Scale; PASAT: Paced Auditory Serial Addition Test; PDQ: Perceived Deficits Questionnaire; PPMS: primary-progressive multiple sclerosis; PRIMuS: Patient Reported Impact of MS; RAVLT: Rey Auditory Verbal Learning Test; RRMS: relapsing-remitting multiple sclerosis; SD: standard deviation; SDMT: Symbol-Digit Modalities Test; SF-36: Short Form-36; SPMS: secondary-progressive multiple sclerosis; SRT: Selective Reminding Test; Stoop: Stroop Color-Word Test; TOH: Tower of Hanoi; UFOV: Useful Field of View Test; UTI: urinary tract infection; WMS: Wechsler Memory Scale; WRAT3: Wide Range Achievement Test 3.

Methods

Single-centre, double-blind, placebo-controlled, randomised clinical trial.

The sample size and power were estimated. On the basis of previous estimates of SD of recall change score 0.9, the researchers calculated that 30 patients per treatment group would be required to provide the study with 90% power to detect (with a 2-sided α of 0.05) a mean difference in general memory score on the WMS from baseline of 2.5.

All patients completed treatment without interruption and were available for follow-up at 12 weeks.

Treatment duration: 12 weeks.

Follow-up period: 12 weeks.

ITT analysis: yes.

Participants

Inclusion criteria: (1) men and women aged 15-55 years with a clinical or laboratory-supported diagnosis of definite MS according to Poser's diagnostic criteria; (2) stable neurological functioning for at least 1 month prior to study entry and have an EDSS score ≤ 6 and a willingness to continue current medications for the duration of the study; (3) displayed at least mild verbal memory impairment as indicated by a WMS; (4) concurrent antidepressants, antispasticity agents, anticholinergic drugs frequently taken for urinary problems, and DMTs were permitted, as long as dosage had been constant for at least 1 month prior to the evaluation; (5) women of childbearing potential had to practice a clinically accepted method of contraception.

Exclusion criteria: (1) patients currently taking benzodiazepines were excluded, as these medications may affect cognition; (2) current alcohol or substance abuse, history of head injury or other medical condition known to affect cognition.

71 patients were eligible for the study. 11 patients were excluded because they refused entry.

60 patients were assigned randomly and equally to 1 of the 2 self administered treatment groups.

Most of characteristics of patients in the 2 treatment groups were generally well matched at baseline except for the subtests of digit span and visual reproduction in WMS (rivastigmine group scored slightly but significantly lower).

Summary of patient characteristics at baseline (placebo: G1 (n = 30), rivastigmine: G2 (n = 30)):

  • Mean age: G1 = 31.6 ± 7.6 years, G2 = 33.4 ± 9.6 years, difference 1.8 years (95% CI -2.7 to 6.2)

  • Gender: men G1 = 12 (40.0%), G2 = 15 (50.0%), difference 10.0% (95% CI -15.0% to 35.0%);women: G1 = 18 (60.0%), G2 = 15 (50.0%)

  • Duration of MS: G1 = 4.5 ± 3.2 years, G2 = 6.3 ± 4.1 years, difference 1.8 years (95% CI -0.8 to3.7)

  • EDSS: G1 3.9 (1.1), G2 4.0 (1.2), difference0.1 (95% CI -0.5 to 0.7)

  • Education: primary or below G1 = 7 (23.3%), G2 = 11 (36.7%), difference 13.3% (95% CI -9.6% to 36.3%); secondary: G1 = 13 (43.3%), G2 = 11 (36.7%), difference 6.7% (95% CI -31.4% to 18.1%); post-secondary: G1 = 10 (33.3%), G2 = 8 (26.7%), difference -6.7% (95% CI -29.8% to 16.5%)

  • Marital status: married: G1 = 19 (63.3%), G2 = 19 (63.3%), difference 0.0% (95% CI -24.4% to 24.4%); single: G1 = 11 (36.7%), G2 = 11 (36.7%)

  • Score on the Wechsler Memory Scale Subtests: information: G1 = 5.8 ± 0.4, G2 = 5.6 ± 0.6, difference -0.2 (95% CI -0.4 to 0.1); orientation: G1 = 4.9 ± 0.3, G2 = 4.9 ± 0.3, difference 0.0 (95% CI -0.1 to 0.1); mental control: G1 = 7.2 ± 1.0, G2 = 7.6 ± 1.0, difference 0.4 (95% CI -0.1 to 0.9); logical memory: G1 = 15.2 ± 2.5, G2 = 16.1 ± 1.8, difference 0.9 (95% CI -0.2 to 2.0); digit span: G1 = 5.0 ± 1.0, G2 = 4.6 ± 0.7, difference -0.9 (95% CI -1.3 to -0.5); visual reproduction: G1 = 9.7 ± 1.4, G2 = 8.1 ± 1.0, difference -1.6 (95% CI -2.2 to -1.0); associative learning: G1 = 12.0 ± 2.0, G2 = 13.0 ± 2.8, difference 1.0 (95% CI -0.3 to 2.2); general memory score: G1 = 60.5 ± 4.9, G2 = 60.0 ± 4.2, difference -0.5 (95% CI -2.9 to 1.9)

  • MS subtypes: RRMS: G1 = 11 (36.7%), G2 = 9 (30.0%), difference -6.7 (95% CI -30.5 to 17.1); SPMS: G1 = 16 (53.3%), G2 = 15 (50.0%), difference -3.3 (95% CI -28.6 to 21.9); PPMS: G1 = 3 (10.0%), G2 = 6 (20.0%), difference 10.0 (95% CI -7.9 to 27.9)

  • Concomitant medications: interferon-β: G1 = 22 (73.3%), G2 = 19 (63.3%), difference -10.0 (95% CI -33.4 to 13.4); antidepressant: G1 = 29 (96.7%), G2 = 25 (83.3%), difference -13.3 (95% CI -28.1 to 1.5); bladder drug: G1 = 19 (63.3%), G2 = 19 (63.3%), difference 0.0 (95% CI -24.4 to 24.4); antispasmodic: G1 = 20 (66.7%), G2 = 22 (73.3%), difference 6.7 (95% CI -16.5 to 29.8).

Interventions

Experimental intervention: rivastigmine 1.5 mg once daily increment over 4 weeks to 3 mg twice daily for a total of 12 weeks (G2).

Control intervention: placebo (G1).

Outcomes

Primary outcome: general memory score on the WMS and its subtests.

Secondary outcome: frequency of AEs.

Notes

60 patients who sought treatment for MS at a neurology clinic in Al-Zahra Hospital affiliated to Isfahan University of Medical Sciences, Iran, between October 2005 and February 2007.

This is the first study that assessed effectiveness of rivastigmine in treatment of cognitive dysfunction in MS.

No information on depression and fatigue at baseline was reported.

Risk of bias
BiasAuthors' judgementSupport for judgement
Random sequence generation (selection bias)Low risk

Quote: "Patients were randomized according to a preexisting list produced by a computer program that differed from a random number generator only in that it assigned equal numbers of patients into each treatment group."

Comment: sequence generation was adequate.

Allocation concealment (selection bias)Low risk

Quote: "The hospital pharmacist was informed of all randomization assignments and was responsible for labelling the study drug and maintaining a master list linking the patients and their treatment assignments."

Comment: allocation concealment was adequate

Blinding (performance bias and detection bias)
All outcomes
Low risk

Quote: "The trial was double-blinded in that both patient and physician who assessed the outcome were not aware of treatment type that the patient was received. The masking of the active and placebo treatments was preserved by creating treatments that looked identical. Patients were evaluated at baseline and 12 weeks after the start of the therapy by a qualified neurologist to evaluate the development of side effects of the medications, compliance of the patients, and efficacy parameters. Efficacy assessments of all patients were administered by a single rater (NZ) in the same sequence, and did not know which patients had received which treatment."

Comment: participants, personnel and outcomes assessors in this study were blinded to the allocated interventions.

Incomplete outcome data (attrition bias)
All outcomes
Low risk

Quote: "The 60 patients who completed treatment were available for follow-up at 12 weeks."

Comment: outcome data were complete.

Selective reporting (reporting bias)Low riskAll listed outcomes were reported adequately.
Other biasUnclear risk

Most of characteristics of patients in the treatment groups were generally well matched at baseline except for the subtests of digit span and visual reproduction in WMS (rivastigmine group scored slightly but significantly lower). No information on depression at baseline was reported.

Comment: this may have led to biased results.

Characteristics of excluded studies [ordered by study ID]

StudyReason for exclusion
  1. a

    MS: multiple sclerosis.

Aragona 2009Cross-over study with administration duration < 12 weeks.
Barak 1999Without memory scales in outcomes.
Barak 2002(1) Non-randomised study; (2) control group were untreated MS patients.
Benedict 2008Administration duration < 12 weeks.
Bever 1996Cross-over study with administration duration < 12 weeks.
Bosca 2004The participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Bruce 2012Cross-over study with administration duration < 12 weeks.
Cabrera-Gomez 2003Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Christodoulou 2006Based primarily on a study previously published that was included in this review.
Cohen 1989Cross-over study with administration duration < 12 weeks.
Cohen 2002Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Cohen 2010Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Comi 2012Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Decoo 2004Cross-over study.
Fischer 1994Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Fischer 2000Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Flechter 2007Non-randomised and open-label study without control group.
Geisler 1996Single-blind study with administration duration < 12 weeks.
Greene 2000Non-randomised, non-controlled, open-label study.
Harel 2009Administration duration < 12 weeks.
Havrdova 2006Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Honarmand 2011Cross-sectional study.
Huolman 2011(1) Non-randomised study; (2) control group were healthy people.
Iaffaldano 2012Non-randomised, non-controlled, open-label, observational cohort study.
Jeffery 2011Without memory scales in outcomes.
Kappos 2010Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Krause 2012(1) Non-randomised study; (2) control group were healthy people.
Krupp 1999Open-label trial.
Lacy 2013Non-randomised, open-label study.
Lang 2012Non-randomised, non-controlled, open-label, prospective cohort study.
Lange 2009Administration duration < 12 weeks.
Mattioli 2011a(1) Non-randomised and single-blind (assessor-blind) study; (2) quasi controls (MS patients administering immunomodulatory drugs).
Mattioli 2011bNon-randomised, non-controlled, observational study.
Melanson 2010Non-randomised, open-label study.
Miller 2011Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Montalban 2009Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Morrow 2009Administration duration < 12 weeks.
Morrow 2013Administration duration < 12 weeks.
Möller 2011Administration duration < 12 weeks.
Oliveri 1998(1) Non-randomised study; (2) control group were healthy people.
Panitch 2004Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Patti 2009Non-randomised, non-controlled, open-label, observational cohort study.
Patti 2010Non-randomised, non-controlled, open-label, observational cohort study.
Pliskin 1994Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Pliskin 1996Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Portaccio 2013Non-randomised study.
Rorie 2001Cross-over study.
Rossini 2011Cross-over study.
Rudick 2006Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Sailer 2000Cross-over study.
Schröder 2011Observational study with an open-label retrospective analysis.
Schwid 2007Open-label trial.
Selby 1998(1) Non-randomised study; (2) control group were healthy people.
Selmaj 2012Without memory scales in outcomes.
Smits 1994Cross-over study with administration duration < 12 weeks.
Sumowski 2011Based primarily on a study previously published that was excluded in this review.
Uttner 2005(1) Administration duration < 12 weeks; (2) untreated control group consisted of healthy people.
Villoslada 2009Cross-over study.
Wade 2004Administration duration < 12 weeks.
Weinstein 1999Participants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Weinstock-Guttman 2012bParticipants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Weinstock-Guttman 2012aParticipants did not meet the inclusion criteria for memory impairment prespecified in the protocol.
Wilken 2004Not a double-blind study.
Wilken 2008Not a double-blind study.
Zéphir 2005(1) Non-randomised study; (2) control group were healthy subjects.
Zéphir 2008(1) Non-randomised study; (2) control group were healthy subjects.

Characteristics of studies awaiting assessment [ordered by study ID]

NCT01074619

  1. a

    DRS: Depression Rating Scale; EDSS: Expanded Disability Status Scale; IV: intravenous; MADRS: Montgomery-Asberg Depression Scale; MRI: magnetic resonance imaging; MS: multiple sclerosis; PASAT: Pace Auditory Serial Addition Test.

MethodsStudy on Cognitive Disorders of Multiple Sclerosis.
Participants

Inclusion criteria:

(1) Aged 18-60 years, both genders

(2) Remitting multiple sclerosis defined by McDonald 2001

(3) Patient with authorised immunomodulator treatment or oral immunosuppressive therapy for > 3 months: interferon-β, glatiramer acetate, azathioprine, methotrexate, mycophenolate mofetil, treatment by monoclonal antibody IV or anti-VLA4, natalizumab (Tysabri)

(4) Patient having benefited, possibly, of following treatments: mitoxantrone, cyclophosphamide, cyclosporine, general-purpose immunoglobulins, only if the treatment is ended > 6 months before the inclusion

(5) EDSS score ≤ 5.5 and DRS score ≥130

(6) PASAT 3s score > 15 and < median/control subjects according to 2 age brackets, gender, school level

(7) Signed the informed consent form

(8) Effective contraception for women in age to procreate

Exclusion criteria:

(1) Progressive form MS

(2) MS relapse of less of 4 weeks

(3) IV or oral corticoid treatment in the month preceding the screening

(4) Medicinal treatments or non-medicinal in cognitive or psychology-stimulant aim in the 3 months before the screening

(5) Tumoural form MS visible by MRI

(6) Depressive syndrome (MADRS score >19)

(7) Quite other diagnosed psychiatric pathology

(8) Known allergy or quite contraindication in memantine: renal or hepatic insufficiency, turned out epileptic disease, treatment by ketamine, amantadine, dextromethorphan, L-dopa, dopaminergic agonist, barbiturates, neuroleptic, 3,4-diaminopyridine, lithium, cimetidine, ranitidine, procainamide, quinine, nicotine, hydrochlorothiazide and ally, phenytoin, modafinil

(9) Recent treatment (< 4 weeks) by antidepressants, anxiolytics, or both

(10) Pregnancy or breastfeeding

(11) Minor or major "protected by the law" patient

(12) Uncontrolled diet

(13) Patient having benefited of 1 psychometric assessment (including in particular tests planned in the protocol) within 1 year.

Interventions

Intervention group: memantine, 5 mg the first week, then 10 mg the second week, 15 mg the third week and 20 mg the fourth week until the end of the study (t0 + 1 year).

Control group: placebo, 5 mg the first week, then 10 mg the second week, 15 mg the third week and 20 mg the fourth week until the end of the study (t0 + 1 year).

OutcomesPrimary outcome measures: PASAT.
Notes

Official title: Effects of Memantine on Cognitive Disorders of Relapsing-remitting Multiple Sclerosis.

This trial started in September 2005 and was completed in November 2011, but we could not find a formal publication. We wrote to the principal investigator requesting information about this trial, but we received no reply.

Contact information: Defer Gilles, Professor, 02.31.06.46.20 ext +33, Email address: defer-gi@chu-caen.fr.

Sponsor: University Hospital, Caen, Collaborators: Ministry of Health, France, H. Lundbeck A/S.

clinicaltrials.gov/ct2/show/NCT01074619.

Characteristics of ongoing studies [ordered by study ID]

NCT01466114

  1. a

    EDSS: Expanded Disability Status Score; HIV: human immunodeficiency virus; MRI: magnetic resonance imaging; MS: multiple sclerosis; PASAT: Paced Auditory Serial Addition Test; PO: orally; QD: once a day; SDMT: Symbol-Digit Modalities Test.

Trial name or titleEstriol Treatment in Multiple Sclerosis (MS): Effect on Cognition.
MethodsRandomised, double-blind (participant, investigator, outcomes assessor), parallel assignment.
Participants

Inclusion criteria:

(1) Diagnosis of clinically definite or MacDonald criteria relapsing-remitting multiple sclerosis, secondary-progressive multiple sclerosis or primary-progressive multiple sclerosis

(2) No relapse within 30 days before day of trial enrolment (month 0 visit). If steroids given for relapse, then the month 0 visit must be 30 days after last steroid dose.

(3) Females aged 18-50 years, and EDSS = 0.0-6.0

(4) Screening PASAT (3-second) score 25-50, inclusive

(5) Must be mentally competent enough to comply with study guidelines and give informed consent and must be willing and able to travel to the study centre at frequencies in the protocol for a total period of 12 months

(6) Patients must be on a stable dose of 1 of the following agents for a minimum of 3 months prior to the month 0 visit: Copaxone®, Betaseron® (or Extavia®), Rebif®, or Avonex®, Gilenya®, Aubagio® or BG-12. The time spent in the screening period may serve as part of this 3-month period

(7) Patients who are currently being treated with adrenocorticotropic hormone, corticosteroids, intravenous immunoglobulins, plasma exchange, Lipitor® or minocycline may be included

(8) Patients who are not currently treated with 1 of the above treatments may only be included if they plan to start Copaxone® or an interferon (Betaseron® (or Extavia®), Rebif®, Avonex®) or an oral agent (Gilenya®, Aubagio® or BG-12) and then they must be on for at least 3 months prior to month 0 (as above).

Exclusion criteria:

(1) Males

(2) Participant on oral contraceptives, hormone replacement therapy, progesterone intrauterine devices or other sex hormones during screening and during the 12-month study period

(3) Females who are pregnant or who plan to become pregnant during the 12 months of enrolment, who wish to become pregnant within 3 months following completion of the study, or who will be within 6 months postpartum at the day of first enrolment visit (month 0)

(4) Females who plan to breastfeed after first enrolment visit (month 0)

(5) Fertile sexually active women who are unwilling to practice reliable barrier methods of contraception other than oral contraceptives

(6) Patients with surgical ovariectomy with no hormone replacement for ≥ 1 year

(7) Menopause with no hormone replacement for ≥ 3 years prior to the first enrolment visit

(8) Patients who smoke at any time during screening or during the 12-month study period

(9) Patients who have serious pulmonary, renal, gastrointestinal, hepatic, immunological, infectious, neoplastic, major psychiatric disease (major depression, schizophrenia), endocrine disease (including major diabetes, thyroid disease), or gynaecological disease

(10) Vitamin B12 level < 200 pg/mL or drug abuse within the past 5 years

(11) > 130% or < 80% of their ideal body weight based on Metropolitan Life Tables

(12) Conditions that would interfere with assessing neurological functions such as deforming arthritis or a major amputation

(13) Have at any time been treated with total lymphoid irradiation, monoclonal antibody, T-cell vaccination, cladribine, bone marrow transplantation, azathioprine, cyclophosphamide, methotrexate, mitoxantrone, cyclosporine, Tysabri

(14) Positive titres to HIV in the past

(15) Previous serious adverse effects with oestrogen treatment

(16) Patients with MS-disease duration (since onset of symptoms) of > 15 years.

Interventions

Experimental group (group A): estriol (Synapause), 2 capsules of 2 mg (total of 8 mg) PO QD; norethindrone (progestin), starting at month 6, and at months 9 and 12: women who are on estriol (group A) take 0.7 mg PO QD for 2 weeks.

Placebo comparator (group B): placebo, 4 capsules PO QD; progestin placebo, starting at month 6 and at months 9 and 12: women who are on placebo (group B) take a second progestin placebo pill PO QD for 2 weeks.

Outcomes

Primary outcome measures: change from baseline in cognitive function assessed by PASAT.

Secondary outcome measures:

(1) Change from baseline in cognitive function as assessed by cognitive evoked potentials, measured in milliseconds

(2) Change from baseline in standard MS outcome measures (relapses, EDSS, 25-foot walk test, 9-hole peg test, low contrast visual acuity, MS Quality of Life, Modified Fatigue Impact Score, Beck Depression Inventory)

(3) Determine safety by assessing the number of women with adverse events with combination treatment compared with placebo

(4) Change from baseline in MRI measurements of whole brain volume, T2 lesion volume and number of enhancing lesions. Whole brain atrophy, T2 lesion volume, gadolinium enhancing lesion number and volume will be assessed on brain MRI at baseline and conclusion in each woman

(5) Change from baseline in cognitive function as assessed by a brief battery of cognitive tests (Processing speed: SDMT; Visual Memory: 7/24 Spatial Recall Test, Benton Forms F & G; Verbal memory: Buschke Selective Reminding Test, Verbal Paired Associates; Language: Word List Generation.

Starting dateOctober 2011.
Contact informationContact: Jenny Bardens, (310)206-2176, jbardens@mednet.ucla.edu; Janet Yau, (310)794-4020, jyau@mednet.ucla.edu.
Notes

Official Title: A Double-Blind, Placebo Controlled Trial of Estriol Treatment in Women With Multiple Sclerosis: Effect on Cognition.

clinicaltrials.gov/ct2/show/NCT01466114.

Ancillary