Genetic polymorphisms and post-stroke upper limb motor improvement – A systematic review and meta-analysis

Introduction

Stroke continues to be a leading cause of adult morbidity in the United States. ¹ One of the most disabling aftereffects of a stroke is the presence of upper limb (UL) hemiparesis. A large proportion of stroke survivors present with UL sensorimotor impairments on the paretic side, reduced independence in performance of daily life activities (ADL) and restricted participation. ² Along with spontaneous recovery mechanisms, ³ motor improvement of the paretic side enabling successful task-performance is attributable to adaptive neuroplasticity and motor learning. ⁴

Successful task-performance entails an interaction of the individual, environment, and the task to be performed. ⁵ The role of the environment^6,7 and intervention-related factors influencing task-practice ⁸ have been extensively studied. Recently, there is a renewed focus on the role of individual-specific characteristics such as levels of motivation,^9,10 mood ¹⁰ and the role of biomarkers. ¹¹ Bernhardt et al ¹² defined biomarkers as “indicators of disease state that can be used clinically as a measure reflecting underlying molecular and cellular processes that may be difficult to measure directly in humans and could be used to predict recovery or treatment response.” Biomarker studies within the realm of neurorehabilitation include those based on biology (eg, genetics), structural and/or functional imaging ¹³ and neurophysiological markers ¹⁴ of central nervous system excitability and electrical activity.

The role of imaging-based biomarkers of structural and functional corticospinal tract connectivity alone ¹³ or in combination with neurophysiological markers (eg, motor evoked potential amplitude) ¹⁴ has been extensively studied. The role of genetics-based biomarkers is slowly gaining prominence, ¹⁵ with studies focusing on single nucleotide polymorphisms (SNPs). ¹¹ These SNPs alter the basic functioning in cellular and molecular processes ¹⁶ and can influence functional improvement produced by (i) environmental interaction and (ii) in response to rehabilitation interventions. ¹⁷ Genetics-based biomarkers pertinent to stroke recovery include SNPs in Brain Derived Neurotrophic factor (BDNF), Catechol-o-Methyltransferase (COMT) and Apolipoprotein (APOE). ¹¹

An activity dependant ¹⁸ neurotrophin, BDNF is important for neuroplasticity and protection after injury. It facilitates synaptic transmission and long-term potentiation important for motor learning. ¹⁹ A common SNP that occurs in BDNF is substitution of 1 or 2 valines at codon 66 (rs6265) with methionine (val⁶⁶met or met⁶⁶met) due to substitution of adenine in place of guanine at nucleotide 196. ²⁰ This polymorphism reduces activity-dependent BDNF release, ²¹ and results in altered learning and neuroplasticity in healthy controls ²² and after a stroke.^23,24

The COMT enzyme helps degrade and thus influences the availability of Dopamine in the central nervous system. ²⁵ Dopamine can influence post-stroke motor learning and improvement.^26,27 A commonly observed SNP (rs4680) results in a substitution from valine or methionine at codon 158 (in the membrane form) and codon 108 in the soluble form. This results in a 3-4-fold decrease in COMT activity.^28,29 The role of COMT polymorphism has primarily been assessed on motor learning in Parkinson’s disease^30,31 and severe Schizophrenia. ³² Given that COMT is found in areas essential for motor learning, ³³ such as striatum and motor cortex, ³⁴ the effects of COMT polymorphism on post-stroke motor improvement need to be addressed.

Although involved in lipid transport between cells, APOE helps modulate neuronal repair and regeneration of nervous tissue. One of the alleles of APOE is the Epsilon-4 form (ε4) with arginine at positions 112 and 158 in place of cystine (rs429358 and rs7412). Presence of APOE-ε4 can cause reduced hippocampal volume and cortical thickness, cognitive impairments ³⁵ and lower recovery levels after traumatic brain ³⁶ and spinal cord ³⁷ injuries. After a stroke, previous meta-analyses^38,39 on the effects of APOE-ε4 revealed lower improvement after sub-arachnoid hemorrhage in those with, compared to those without the ε4 form. However, no association was reported with improvements noted after ischemic strokes. In both studies,^38,39 motor improvements were assessed using generic scales such as the modified Rankin Scale (mRS). Improved scores in assessments such as the mRS does not specifically represent UL motor improvement. ⁴⁰ As presence of cognitive impairments influence UL motor improvement, ⁴¹ the effects of the APOE-ε4 form on post-stroke UL motor improvements needs to be systematically evaluated.

The influence of polymorphisms in BDNF and APOE on global stroke recovery has previously been reviewed.^{11,16,24,39,42,43} These studies were either narrative reviews^11,16,24,43 or meta-analyses including global stroke outcomes like National Institutes of Health Stroke Scale and/or mRS.^39,42 Post-stroke UL motor improvement continues to remain variable and less than optimal in many cases. ⁴⁴ Evaluation of whether and to what extent genetic polymorphisms influence the extent of improvement may help explain some of the observed variability. Using a systematic review and meta-analysis, we examined the influence of genetic polymorphisms on UL motor improvement. The question guiding our review was “In individuals with post-stroke UL hemiparesis, does the presence, compared to the absence of genetic polymorphisms, influence motor improvement?” Preliminary results have previously appeared as an abstract. ⁴⁵

Methods

This systematic review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) guidelines. The protocol was registered on the Open Science Framework (https://osf.io/wk9cf/).

We searched the literature for studies involving human subjects published in English between the years 2000 and 2023. The last search was conducted in November 2023. Key search terms used included: stroke, cerebrovascular accident, upper limb, arm, rehabilitation, impairment, activities of daily living, recovery, polymorphisms, gene*, neuroplasticity, and motor learning. The * sign after the word gene was used as a wildcard symbol which helped in searching for different words starting with gene including gene, genes, genetic and genetics. Databases searched included: PubMed and ISI Web of Science and the Google Scholar repository. We included studies that used clinical assessments of UL motor impairment and/or ADL and provided data for individuals with and without polymorphisms. We excluded studies focusing exclusively on lower limb or on only cognitive outcomes. We also excluded other reviews, although we searched the reference lists of these excluded reviews for pertinent citations. To identify additional relevant articles, we also searched reference lists of each retrieved study. The searches were initially conducted by 3 authors (RTM, CR and KMS). A separate search was run by the first author (SKS) at the same time, which was repeated again before the manuscript was submitted for publication. No major differences were found in these searches.

Results

The search and selection results are shown in the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) flow diagram. In total, 319 citations were identified through database and registry searches (Figure 1). After removing duplicates, 132 citations were screened, of which 16 were excluded. We sought 116 reports for retrieval and assessed 31 for full text eligibility, which were experimental studies including outcomes related to rehabilitation. We further excluded 21 studies, as they included lower limb and/or gait outcomes or used generic measures such as the mRS, NIHSS and Barthel Index. Ten articles assessing the effects of genetic polymorphisms on UL motor impairment and ADL performance were included in the qualitative synthesis (Figure 1). The reference lists of these ten articles did not yield any additional citations. These studies compared the differences between the dominant (BDNF: val⁶⁶val; COMT: val¹⁵⁸val; APOE: ε4 negative) and co-dominant (BDNF: val⁶⁶met; COMT: val¹⁵⁸met; APOE: 1 ε4) or recessive (BDNF: met⁶⁶met, COMT: met¹⁵⁸met; APOE: both ε4 forms present) genotypes.OPEN IN VIEWER

Out of the 10 studies, 8 examined the influence of the presence of 1 or both met alleles in BDNF and 2 addressed the effects of 1 or 2 met alleles in COMT. Two of the 8 studies assessing the effects of 1 or 2 met alleles in BDNF additionally examined effects of the presence of the APOE ε4 isoform. Six^23,57-61 of the 8 articles addressing effects of 1 or 2 met alleles in BDNF had available FMA scores assessed at the end of the intervention to be used for a meta-analysis. Two of these studies^59,60 also included a retention assessment, with that data being included for a second meta-analysis.

BDNF polymorphism (val⁶⁶met and/or met⁶⁶met)

In total, 598 individuals (59.2% men, 40.8% women) sustaining a stroke participated in the 8 studies included in the qualitative analysis. The average age of the participants (mean ± SD) was 58.4 ± 3.2 years. A greater proportion of participants had sustained ischemic strokes (79.7%) compared to hemorrhagic strokes (20.3%). The distribution of the more-affected side was almost equal (50.7% right, 49.3% left). Three^57,61,62 of the included studies were ranked as ‘good’ and the remaining five^23,58-60,63 ‘fair’ (Supplemental Table 1). Participants were either in the acute^57-60,63 or chronic^23,61,62 stage post-stroke. All participants had moderate-to-severe ⁶⁴ UL motor impairment (FM score ≤49/66).

Table 1 presents a summary of studies evaluating the effects of BDNF val⁶⁶met and met⁶⁶met polymorphism with a focus on sample size, type and dose of rehabilitation provided (if any), main outcomes and results. The sample size used for the Meta-analysis was 295 (no polymorphism: 101, polymorphism: 194). Analysis revealed a large (0.50, 95% CI: 0.11 - 0.88, P = 0.01, I² = 54%, random effects model; Figure 2) effect size at the end of the intervention period for improvement in UL FMA scores in those without compared to those with the polymorphism. At retention testing, the sample size used was 79 (no polymorphism: 19, polymorphism: 60). We found a similar large effect size (0.58, 95% CI: 0.06 - 1.11, P = 0.03, I² = 0%, random effects model; Figure 3).

OPEN IN VIEWER

Table 1.

Effect of BDNF val⁶⁶met and/or met⁶⁶met polymorphism.

Study; Sample size (n); Val/Val and Met Allele distribution and Down’s and Black score	Intervention	Rehabilitation provided/Dose	Outcomes and timing of assessment	Results
Chang et al, 2014; n = 44; Val/Val: n = 9; Met allele: n = 35 DBS: 19 (good)	10 sessions of rTMS over 2 weeks. Each session had 50 trains of 10 Hz frequency for 5 seconds at 90% RMT	Each train of rTMS was followed by 50 seconds of reaching and grasping exercises.	• Upper and lower limb Fugl-Meyer Assessment (FMA) scores • Box and Blocks Test (BBT). Assessments conducted at baseline, end of the intervention and 2-month retention.	FMA
		Active and active assisted exercises consisting of range of motion exercises, moving, and grasping and releasing cups and cubes.		• Upper limb: Greater change seen in Val/Val group at post (10 points) and retention (23 points) compared to Met alleles (4 and 11 points) respectively (P < 0.05).
		All participants also received conventional Physical (PT) and Occupational Therapy (OT) sessions, involving gait, fitness, and ADL training for 3 hours each day.		• Lower limb: Both groups improved at both assessments with no between group difference.
				BBT:
				• Greater change seen in Val/Val group (16 blocks) compared to Met alleles (6 blocks; P < 0.05) at retention.
Chang et al, 2016; n = 62; Val/Val: n = 12 Met allele; n = 50 DBS: 18 (fair)	10 sessions of rTMS over 2 weeks. Each session had 20 trains of 50 stimuli of 10 Hz frequency at 90% RMT	Each train of rTMS was followed by 50 seconds of reaching and grasping exercises.	• Upper and lower limb and total FMA scores	• 20 participants were good responders and 42 were poor responders.
		All participants also received conventional PT and OT sessions, involving gait, fitness, and ADL training for 3 hours each day.	• Degree of preserved Corticospinal Tract (CST) integrity quantified by diffusion tensor imaging and presence/absence of MEP in the FDI muscle.	• Greater proportion of good responders had Val/Val genotype (35%) compared to poor responders (11%).
			Assessments conducted at baseline and end of the intervention.	• Those with Val/Val genotype had significantly higher change in upper limb FMA scores (13.7 points) compared to Met alleles (1 point)
			Patients classified as good or poor responders depending upon the amount of change in UL FMA scores: Good responders (≥5 points); poor responders (≤4 points)	• Individuals with Val/Val genotype almost twice more likely to have better improvement than those with Met alleles.
Kim et al, 2016a n = 35; Val/Val: n = 10 Met allele; n = 25 DBS: 17 (fair)	No details provided	No information provided	• Upper limb FMA scores	UL FMA scores
			• Values of fractional anisotropy (FA), axial diffusivity (AD) and radial diffusivity (RD) for CST.	• Greater change in the Val/Val group at T2 (7 points) and T3 (17 points) compared to the Met group (4 and 12 points respectively).
			Assessments conducted at baseline, T1 (1 month after baseline) and T3 (3 months after baseline)	• Moderate positive correlation with FA values at T1 (r = 0.78) and T2 (r = 0.72) for the Val/Val group and at T3 (r = 0.59) for the Met group
				• Moderate positive correlation with AD scores at T1 (r = 0.78) and T2 (r = 0.72) for the Val/Val group.
				• Moderate negative correlation with RD values scores at T3 (r = −0.59) for the Met group.
Kim et al, 2016b n = 42; Val/Val: n = 26 Met allele; n = 16 DBS: 16 (fair)	Robotic therapy spread over 2-3 weeks.	Therapy consisted of repeated grasp and release movements of the affected hand and wrist.	• Upper limb FMA scores	• No difference in change in upper limb FMA scores between the Val/Val group at T2 (2.1 points) and the Met group (3.2 points) at the end of therapy.
		Participants wore a Hand Wrist assistive rehabilitation device and practiced tasks with real objects as well as virtual reality games (eg, squeezing lemons, moving jewels into a safe, etc.)	• Values of percentage change on fMRI signals and activation volume obtained from the ipsi- and contralesional primary sensorimotor cortex and dorsal premotor cortex using fMRI.	• However, the Val/Val group had greater percentage signal change (P = 0.037) and activation volume (P = 0.03) in the ipsilesional primary sensorimotor cortex compared to the those in the Met group.
		Tasks focusing on speed, reaction time, force, and range of motion were also practiced.	Assessments conducted at baseline, T1 (end of therapy)
Shiner et al, 2016 n = 54; • Val/Val: n = 27 • Met allele: n = 27 DBS: 19 (good)	Wii based movement therapy (n = 40) or modified constraint induced movement therapy (mCIMT, n = 14).	10 one-hour long sessions on consecutive weekdays.	• Upper limb FMA • Wolf Motor Function Test - timed task (WMFT- tt) • Motor Activity Log Quality of Movement (MAL-QoM) scores. Assessments conducted at baseline and at the end of the intervention.	• All participants improved on FMA, WMFT -TT and MALQoM scores at the end of the interventions.
		Sessions targeted movements of the more-affected hand and arm.		• Overall, no difference in amount of change in FMA, WMFT-tt and MAL-QoM scores between those with Val/Val and Met alleles.
		The Wii group played golf, bowling, baseball, boxing, and tennis using the controller in the more-affected arm. mCIMT group received task-based training on object manipulation focusing on movement speed.		• However, subgroup analysis revealed less change in those with Met alleles and moderate (8.9%) or high (13.8%) functional ability on WMFT-tt scores compared to those with Val/Val (25.2% and 37.3% respectively).
		Individuals classified into those with low (inability to move >1 block on BBT), moderate (inability to complete Perdue Pegboard test) and high (ability to complete Perdue Pegboard test) functional ability.		• Similar results were seen on FMA values with  less change in those with Met alleles and moderate (7%) or high (2%) functional ability compared to those with Val/Val (10% and 4% respectively).
Chang et al, 2017; n = 97; Val/Val: n = 21 Met allele: n = 76 DBS: 14 (fair)	Traditional inpatient rehabilitation	Traditional inpatient (2 hrs PT, 1 hr OT) followed by outpatient rehab (1 hr PT, 30 mins OT) or home exs.	• Upper limb FMA scores.	Individuals with normal FMA scores or mild and moderate impairment
			Assessments conducted at baseline and T1 (after 3 months).	• Baseline FMA scores explained 47% of variance in FMA scores at T1
			Participants classified into 4 categories based upon FMA scores at T1: Normal: 66, mild impairment (41-65), moderate impairment (25-40), and severe 0-24).	Individuals with severe motor impairment
				• A combination of presence of Met alleles, baseline FMA scores and age explained 59.5% of the variance in FMA scores at T1. Individuals with Met alleles were 1.48 times less likely to have better scores on the upper limb FMA.
				• Smaller proportion of individuals with two (10%) or one Met allele (31%) recovered significantly at T1compared to those with Val/Val genotype (42.9%).
				• Significant correlation between number of Met alleles and FMA score at T1 (rho = −0.248, P < 0.05).
Park et al, 2020; DBS: n = 58; Val/Val: n = 17 Met allele: n = 41 17 (fair)	Traditional inpatient rehabilitation	All participants received the same dose of PT and OT (3-week intensive inpatient rehabilitation).	• Upper limb FMA scores.	Upper limb FMA scores
			• FA values for CST, intrahemispheric connection from M1 to the ventral premotor cortex and corpus callosum (CC).	• Similar mean change seen in upper limb FMA scores in those with Val/Val (12.6 points) and Met alleles (13.8 points) at T2.
			Assessments conducted at baseline and T2 (after 3 months).	• In the Val/Val group, moderate negative correlation with FA in contralesional intrahemispheric connection from M1 to the ventral premotor cortex at T2 (r = −0.60; P = 0.024).
				• In those with Met alleles, moderate positive correlation with FA in the in the ipsilesional CST (r = 0.47; P = 0.003) and FA in the CC (r = 0.41, P = 0.011).
Cramer et al, 2022; n = 206; Val/Val: n = 166; Met allele: n = 40 DBS: 21 (good)	Task oriented upper extremity training or OT	Participants were randomized to 30 hrs each of task-oriented upper extremity training (Accelerated skill acquisition program), dose-equivalent occupational therapy, or standard of care.	• Change in Log WMFT-tt. • Cerebral atrophy measured using ventricular brain ratio. Assessments carried out at baseline and at end of 12 months	• Overall, no difference in amount of change in Log WMFT-tt scores between individuals with Met alleles compared to those with Val/Val genotype.
				• Greater cerebral atrophy (P < 0.01) seen in individuals with Met alleles compared to those with Val/Val genotype.
				• This enlargement was caused primarily by an enlargement in ventricular volume (P = 0.0098).

ADL: Activities of Daily Living;DBS: Downs and Black Checklist Score; FDI: Flexor Digitorum Indicis; fMRI: Functional Magnetic Resonance Imaging; MEP: Motor Evoked Potential; rTMS: Repetitive Transcranial Magnetic Stimulation; RMT: Resting Motor Threshold.

OPEN IN VIEWER

Figure 2.

Results of meta-analyses examining influence of genetic polymorphisms on upper limb motor impairment quantified using the Fugl-Meyer Assessment, at the end of the intervention period. Larger squares indicate bigger study effect sizes. The diamonds represent pooled effects of results of individual studies. The location of the diamond indicates the estimated effect size and precision of the estimate is indicated by the width of the diamond.

OPEN IN VIEWER

Figure 3.

Results of meta-analyses examining influence of genetic polymorphisms on upper limb motor impairment quantified using the Fugl-Meyer Assessment, at retention testing. Larger squares indicate bigger study effect sizes. The diamonds represent pooled effects of results of individual studies. The location of the diamond indicates the estimated effect size and precision of the estimate is indicated by the width of the diamond.

In addition to UL FMA scores, other assessments at the body structure and function level included use of functional magnetic resonance imaging (fMRI), diffusion tensor imaging (DTI) and MRI. In terms of fMRI outcomes, lower ipsilesional activation volume and percentage signal change were noted in individuals with the Met alleles as compared to the Val homozygous individuals. ²³ Use of DTI revealed differences in radial and axial diffusion ⁵⁹ and fractional anisotropy ⁶⁰ between individuals with and without Met alleles. Individuals with Met alleles also had greater cerebral atrophy on MRI. ⁶²

Sensitivity analysis included an additional meta-analysis being conducted with data from 5 studies included in this analysis. The only excluded ⁵⁹ study provided no details on whether and if so, how many sessions of any form of rehabilitation were provided to the participants. The sample size used for this meta-analysis was 260 (no polymorphism: 91, polymorphism:169). Analysis revealed a large (0.43, 95% CI: 0.01 - 0.86, P = 0.046, I² = 57%, random effects model; Figure 4) effect size at the end of the intervention period for improvement in UL FMA scores in those without compared to those with the polymorphism.OPEN IN VIEWER

Figure 4.

Results of sensitivity analysis (meta-analyses) examining influence of genetic polymorphisms on upper limb motor impairment quantified using the Fugl-Meyer Assessment, at the end of the intervention. Larger squares indicate bigger study effect sizes. The diamonds represent pooled effects of results of individual studies. The location of the diamond indicates the estimated effect size and precision of the estimate is indicated by the width of the diamond.

APOE ɛ4 polymorphism

Table 2 presents a summary of studies evaluating the effects of APOE ɛ4 and COMT val¹⁵⁸met or met¹⁵⁸met polymorphism. Two (good quality^61,62) of the 8 studies examining the effects of BDNF polymorphism also assessed the effects of APOE ɛ4 polymorphism. These 2 studies included a total of 260 participants (61.5% men, 38.5% women). A greater proportion of participants had sustained ischemic strokes (83.4%) compared to hemorrhagic strokes (16.7%). The distribution of the more-affected side was equal (50 % right, 50% left).

OPEN IN VIEWER

Table 2.

Effect of APOE and COMT polymorphism.

Study; Sample size (n); distribution and Down’s and Black score	Intervention	Rehabilitation provided/Dose	Outcomes and timing of assessment	Results
A. Effects of APOE ε4 polymorphism
Shiner et al., 2016; n = 54; ε4: n = 9; non ε4: n = 45 DBS: 19 (good)	Wii based movement therapy (n = 40) or modified constraint induced movement therapy (mCIMT, n = 14).	10 one-hour long sessions on consecutive weekdays.	• Upper limb FMA • Wolf motor function test - timed task (WMFT- tt) • Motor activity Log quality of movement (MAL-QoM) scores. Assessments conducted at baseline and post-intervention.	• All participants improved on FMA, WMFT-TT and MALQoM scores at the end of the interventions.
		Sessions targeted movements of the more-affected hand and arm.		• Overall, no difference in amount of change in FMA and MAL-QoM scores between ε4 carriers and those with non ε4 genotype.
		The Wii group played golf, bowling, baseball, boxing, and tennis using the controller in the more-affected arm. mCIMT group received task-based training on object manipulation focusing on movement speed.		• ε4 carriers tended to take longer to complete ADL activities (WMFT-tt, P = 0.057) compared to those with non ε4 genotype.
		Individuals classified into those with low (inability to move >1 block on BBT), moderate (inability to complete Perdue Pegboard test) and high (ability to complete Perdue Pegboard test) functional ability.
Cramer et al, 2022; n = 206; ε4: n = 61; non ε4: n = 145 DBS: 21 (good)	Task oriented upper extremity training or occupational therapy	Patients were randomized to 30 h each of task-oriented upper extremity training (Accelerated skill acquisition program), dose-equivalent occupational therapy, or standard of care.	• Change in Log WMFT-tt. • Cerebral atrophy measured using ventricular brain ratio. Assessments carried out at baseline and at end of 12 months.	• Overall, no difference in amount of change in Log WMFT-tt scores between ε4 carriers and those with non ε4 genotype.
				• No differences seen in cerebral atrophy between individuals with and without the polymorphism.
B. Effects of COMT polymorphism (val158met and/or met158met)
Liepert et al, 2013; n = 83, Val/val = 12 met allele: 71 DBS: 19 (good)	Traditional rehabilitation	Rehabilitation program included PT, OT, endurance, and strength training. Program was adapted to individual needs of the patient. Details unavailable on total duration of therapy.	• Rivermead motor assessment (RMA) and Barthel Index (BI) scores. RMA scores divided into gross function, leg and trunk and upper limb function.	RMA scores • Individuals with 2 met alleles showed less improvement in gross function (P = 0.003), leg and trunk function (P = 0.022) as well as upper limb function (P = 0.047). • Significant correlation with BI scores at all time points (P < 0.001).
			Assessments carried out at baseline,4 weeks later and at the end of 6 months.
Kim et al, 2016 n = 74, Val/val = 41 met allele = 33 DBS = 15 (fair)	No details provided	No details provided.	• Total Fugl-Meyer assessment (FMA) and functional independence measure (FIM) total scores.	FMA scores • Lower scores at discharge, 3-mos and 6 mos post discharge assessments in those with met alleles (P < 0.01) compared to the val heterozygous group.
			Assessments carried out at hospital admission, discharge 3- and 6-mos Post- discharge.
				FIM scores • Lower scores at discharge (P < 0.01), 3-mos and 6 mons post discharge (P < 0.05) assessments in those with met alleles compared to the val heterozygous group.

DBS: Downs and Black Checklist Score; OT: Occupational Therapy; PT: Physical Therapy.

Both studies used the Wolf Motor Function Test - timed test (WMFT-tt) as the primary outcome. No differences were noted between individuals with and without the APOE ε4 forms on WMFT-tt scores (Table 2A). In addition, groups did not differ on the amount of change seen in UL FMA scores and self-reported levels of UL quality (assessed using the Motor Activity Log) ⁶¹ or in the amount of cerebral atrophy noted between groups. ⁶²

COMT polymorphism (val¹⁵⁸met and/or met¹⁵⁸met)

Two studies (1 good ⁶⁵ and 1 fair ⁶⁶ quality) including 157 participants (59.7% men, 40.3% women) examined the influence of COMT val¹⁵⁸met or met¹⁵⁸met polymorphism (53: no polymorphism, 104: polymorphism). A greater proportion of participants had sustained ischemic strokes (83.8%) compared to hemorrhagic strokes (16.2%). The distribution of the more-affected side was 46.8% right side, 51.9 %, left side and 1.3% of the participants had bilateral strokes. The studies used either the UL section of the Rivermead Motor Assessment (RMA) ⁶⁵ or the FMA. ⁶⁶

Compared to those with met¹⁵⁸met, participants with val¹⁵⁸met allele (ES = 0.51) or val¹⁵⁸val (ES = 0.76) distribution had greater recovery with large effect sizes on the UL section of the RMA, at the end of the intervention period. ⁶⁵ Individuals with val¹⁵⁸val distribution had greater recovery on the total FMA at the end of the intervention period (ES = 2.69), and at 3 (ES = 1.51) and 6 months (ES = 1.98) retention testing. ⁶⁶ In addition, participants without the polymorphism improved more on other components of the RMA scores ⁶⁵ including gross function, leg, and trunk function and higher Functional Independence Measure (FIM) scores. ⁶⁶ Lack of available data from 2 or more studies using the same outcome measures precluded us from conducting a meta-analysis on the effects of APOE ɛ4 and COMT (val¹⁵⁸met, met¹⁵⁸met) polymorphisms.

Discussion

Results from this systematic review and meta-analysis (on the effects of BDNF val⁶⁶met and met⁶⁶met polymorphism) indicated that the presence of some genetic polymorphisms negatively influence post-stroke UL motor improvement. The meta-analyses revealed that presence of BDNF val⁶⁶met and met⁶⁶met polymorphism negatively impacted UL motor improvement assessed using the FMA, immediately after the end of the intervention period as well as at retention testing. Overall, majority of the studies had low risk of bias, which lends further credence to these results. Sensitivity analyses revealed that results continued to remain significant even with the exclusion of the study where information on some domains of bias was not available.

These results are in agreement to those found previously, ⁴² and extend those findings more specifically to UL motor improvement and not just general recovery from a stroke. We also found that while APOE ɛ4 polymorphism does not influence UL motor improvement, presence of COMT val¹⁵⁸met and/or met¹⁵⁸met polymorphism has a negative impact. Our results for APOE ɛ4 agree partially with those found previously, ³⁹ and go beyond those results by focusing on UL motor improvement. To our knowledge, this is the first review which has systematically investigated the effects of COMT polymorphism on post-stroke UL motor improvement.

Limitations

We only included studies involving adult participants published in English (since no one in the team was proficient in other languages). It might be possible that we missed studies published in other languages. None of the studies had an explicit sample size calculation. Information on baseline levels of depression and/or intake of anti-depression medication was available in only 2 studies.^23,65 Information on the presence of depression is essential, as the presence of genetic polymorphisms is an additional risk factor^84–86 for post-stroke depression and can influence the extent of UL motor improvement. ¹⁰ None of the 4 studies^23,61,62,65 published from outside Asia conducted an analysis on the rates of recovery assessed using clinical and/or radiological outcome measures between different ethnicities. Future studies will also need to address the effects of ethnicity on stroke rehabilitation outcomes.

Conclusion

Results of our review suggest that presence of genetic polymorphisms in BDNF (val⁶⁶met and met⁶⁶met) and COMT (val¹⁵⁸met and/or met¹⁵⁸met) negatively impact post-stroke motor improvement. This was confirmed at the body-structure and function domain of the ICF for the UL using the meta-analysis on the effects of BDNF val⁶⁶met and met⁶⁶met polymorphism. Our findings may contribute to the understanding of 1 of the underlying mechanisms to help explain some variability in post-stroke UL motor improvement. This is valuable information for the means of tailoring a plan of care, creating realistic goals, and providing relevant, individualized rehabilitation to every patient. In addition, new questions have been identified including does the (i) use of a fixed number of repetitions result in similar or better levels of UL motor improvement in individuals with genetic polymorphisms; (ii) presence of COMT val¹⁵⁸met and/or met¹⁵⁸met continue to influence motor improvement at retention testing; (iii) presence of genetic polymorphisms influence participation levels and (iv) do individuals with genetic polymorphisms use altered movement patterns and if so, to what extent. Answers to these emergent questions can help better understand the influence of genetic polymorphisms on post-stroke upper limb motor improvement.

Supplemental Material

Ethical statement