10.1007@s00415 020 09815 2
10.1007@s00415 020 09815 2
10.1007@s00415 020 09815 2
https://doi.org/10.1007/s00415-020-09815-2
REVIEW
Abstract
Background Spinocerebellar ataxias (SCAs) are rare dominantly inherited neurodegenerative disorders that lead to severe
disability and premature death.
Objective To better characterize the natural history of the most common SCAs, SCA1, SCA2, SCA3 and SCA6, we per-
formed a meta-analysis of literature to determine disease progression, provide data for sample-sizes calculations for inter-
ventional trials and study the impact of geographical locations and study follow-up on disease progression.
Methods A systematic literature search from MEDLINE and EMBASE databases for longitudinal natural history studies
of SCA patients was conducted. Studies using the Scale for the Assessment and Rating Ataxia (SARA) as outcome measure
were considered. Random-effect (RE) meta-analysis was applied to estimate pooled disease progression.
Results Six studies with 1215 SCA patients enrolled between 2005 and 2016 were finally selected. Annual pooled SARA
score increase was 1.83 (1.46–2.20) in patients with SCA1, 1.40 (1.19–1.61) in patients with SCA2, 1.41 (0.97–1.84) in
patients with SCA3, and 0.81 (0.66–0.97) in patients with SCA6. For patients with SCA3, disease progression was faster in
studies located in Asia and Europe than in the US. Two-arm interventional trials of 1-year duration to achieve 80% power
and α level of 5% would require 92 patients per group with SCA1, 97 with SCA2, 115 with SCA3, and 430 with SCA6 to
detect a 50% reduction in disease progression.
Conclusion This meta-analysis provides quantitative data on the progression of the most common spinocerebellar ataxias
based on patient numbers that exceed those of previous studies and confirms that disease progression is faster in SCA1,
intermediate in SCA2 and SCA3 and slower in SCA6, with similar rates of disease progression in SCA2 and SCA3 between
different populations, suggesting a possibility of international collaborative studies. Nevertheless, individual-patient data
meta-analysis is needed to better understand the risk factors that influence disease progression and improve patient stratifi-
cation in interventional trials.
Introduction
Alhassane Diallo and Heike Jacobi contributed equally. Spinocerebellar ataxias (SCAs) are inherited rare neurologi-
cal disorders, characterized clinically by progressive coordi-
Electronic supplementary material The online version of this
article (https://doi.org/10.1007/s00415-020-09815-2) contains nation problems that lead to severe disability and premature
supplementary material, which is available to authorized users.
3
* Alhassane Diallo German Center for Neurodegenerative Diseases (DZNE),
alhassane.diallo@inserm.fr Bonn, Germany
4
1 Sorbonne Universités, INSERM U 1136, Institut Pierre Louis
INSERM, U1137, CIC‑EC 1425, Department
D’Epidémiologie et de Santé Publique IPLESP, AP-HP,
of Epidemiology, Biostatistics, and Clinical Research,
Hopitaux Universitaires Pitié-Salpêtrière-Charles Foix,
AP‑HP, Hospital Bichat, University Paris Diderot Paris,
75013 Paris, France
Paris, France
5
2 Department of Neurology, University Hospital of Bonn,
Department of Neurology, University Hospital of Heidelberg,
Bonn, Germany
Heidelberg, Germany
13
Vol.:(0123456789)
Journal of Neurology
death. More than 40 genetically distinct SCAs have been Study selection, participants, and outcome measure
defined with the most common being SCA1, SCA2, SCA3,
and SCA6. These subtypes are caused by abnormal CAG Natural history studies were included if they had a prospec-
repeat expansions, encoding elongated polyglutamine tracts tive cohort design with at least 12 months follow-up and used
within the proteins associated with each type [1]. the SARA score ranging from 0 (absence of ataxia) to 40
Recently, considerable effort has been made in charac- (most severe ataxia) as primary outcome measure to assess
terizing the natural history in each genotype, and identify- the disease progression [10]. Participants aged 15 years or
ing factors that determine disease progression [2–9]. These older of any sex, ethnicity or geographical location (Europe,
studies concluded that the progressive ataxia measured by America, and Asia) and disease severity, who had a positive
the Scale for the Assessment and Rating Ataxia (SARA) molecular genetic test for SCA1, SCA2, SCA3, and SCA6, and
[10] is the leading symptom in these disorders and increases progressive, otherwise unexplained ataxia were considered.
linearly over time, with different rates in each genotype. Pro- Case–control and cross-sectional studies, and cases series or
gression is faster in SCA1, intermediate in SCA2 and SCA3, cases reports were excluded.
and slower in SCA6 [5, 7]. Due to the genetic background
and the length of follow-up, the rate of progression and fac- Data extraction
tors that influence progression varied between studies. These
rates of ataxia progression are comparable between Euro- Data were extracted by two independent reviewers (AD and
pean and American (US and Brazil) populations, but lower HJ), and in case of disagreement, the relevant paper was
than those of Asian populations [9]. reviewed and differences were resolved by consensus. For each
As potential drugs for SCAs are being developed, the natural history study that met inclusion criteria, the following
summarized progression rates of each genotype from dif- information was extracted: author names, year of publication,
ferent populations are needed to assess more accurately the country, study design (follow-up, prospective cohort), sam-
estimated sample size and treatment effects. The aim of this ple size, study population at baseline (age, sex, age at disease
systematic review and meta-analysis was to synthetize avail- onset, disease duration, number of expanded CAG repeats,
able natural history studies of the most common SCA types SARA score), disease subtype, and effect sizes (annual slope
1, 2, 3, and 6, as well as to identify factors that determine progression ± standard error or annual slope progression with
the disease course. 95% confidence interval).
13
Journal of Neurology
variables. Original natural history studies were described from each genotype, respectively. Residual errors (γε)
using the study characteristic summary Table 1 and for- were obtained via a 1000 simulations procedure, and more
est plot. Potential bias was checked graphically (fun- details are given in appendix. All analyses were performed
nel plot) and statistically (Egger’s regression test) [14]. using the R package ‘metafor’; P < 0.05 were considered
Pooled effect sizes for disease progression for each geno- as significant for effect sizes and publication bias.
type were estimated using the random-effects (RE) meta-
analysis with the restricted maximum likelihood estima-
tion (REML), and summarized as mean slope progression
with 95% confidence intervals. To assess the heterogeneity Results
between studies, we used the I2 statistic ranging from 0
(no observed heterogeneity) to 100% (maximum hetero- Screening results
geneity), with I2 greater than 50% indicating a substan-
tial heterogeneity [15]. To examine a potential source Initial systematic search from MEDLINE and EMBASE
of heterogeneity by the geographical location (Europe, identified a total of 145 articles. After removing duplicated
America, and Asia), and the total duration of follow-up, a articles and further screening by title and abstract, there
subgroup analysis using the meta-regression method was were 12 remaining eligible articles for the full review.
performed. We considered a maximum follow-up less than Six articles that did not meet the inclusion criteria were
5 years as a shorter follow-up and more than 5 years as excluded after the full-text review, resulting in six natural
a longer follow-up. Based on the pooled effect sizes of history studies to be included in the final analysis. Figure 1
disease progression for each genotype, we calculated the summarizes the screening and selection processes. The
sample size needed for two-group interventional trials of quality of included studies was overall good ranging from
1-year duration to achieve 80% and 90% power with effect 7 to 8 out of 8 on the Newcastle–Ottawa Quality Assess-
sizes ranging from 20 to 100%. To compute the sample ment Scale (table S1). For studies with seven starts, inap-
size, we approximated the standard deviation SD using propriate statistical models or insufficient justification for
the formula: SD = SEβ + γε, where S Eβ was the standard baseline covariate included in regression models was the
error of the pooled effects sizes from the random-effects most common issue.
meta-analysis, and γε the estimation of the residual error
Data are summarized by mean for continuous variable and number for categorical variables
CAGlength of expanded repeats allele, SARAScale for the Assessment and Rating Ataxia, SCA spinocerebellar ataxia, NR not reported
13
Journal of Neurology
Identification
database searching MEDLINE through other sources (n=0)
(n=127); EMBASE (n=18)
Duplicated removed
(n=13)
Screening
Study and participant characteristics of age at baseline, age at disease onset, and severity of ataxia
differed from one SCA to another (Table 1).
The selected six natural history studies had a longitudi-
nal prospective cohort design including 1215 patients Disease progression
with the most common types of SCAs (SCA1 = 202,
SCA2 = 342, SCA3 = 436, and SCA6 = 235). Among them, Applying the random-effects (RE) meta-analysis, the pooled
48% were enrolled in Europe, 32% in the US and 19% in rate of annual increase of total SARA score of patients with
Asia between 2005 and 2016 at 48 ataxia referral centers SCA1 was 1.83 (95% confidence interval 1.46–2.20). The
around the world (24 in Europe, 12 in US, 8 in Japan, 3 corresponding annual increases of SARA score was 1.40
in Brazil, and 1 in Taiwan), and had a mean observation (1.19–1.61) in patients with SCA2, 1.41 (0.97–1.84) in
time of 4.3 years [range 2–8 years]. The included studies patients with SCA3, and 0.81 (0.66–0.97) in patients with
were published between 2011 and 2018 with a median SCA6 (Fig. 2). Except for patients with SCA3 (I2 = 89%;
impact factor at the publication moment of 4.505 [IQR p < 0.0001), we did not find a significant within-disease
3.783–5.864]. subtype heterogeneity. To compare disease progression
Overall, the disease subtype SCA3 was the most prevalent across genotypes, we applied a subgroup analysis using a
with 35.9% of the sample, followed by the SCA2 (28.1%), meta-regression method. We found that for any SCAs, the
SCA6 (19.3%), and SCA1 (16.6%). Irrespective of the geno- annual disease progression increases by 1.32 (1.10–1.55).
type, more than the majority of patients were men. The range As expected, this progression was associated with large
13
Journal of Neurology
Fig. 2 Forest plot of the ataxia disease progression. SCA spinocere- p for heterogeneity. Disease progressions with 95% confidence inter-
bellar ataxia, RE random-effect, I2 quantify heterogeneity, τ2 magni- val were calculated using a random-effects meta-analysis with the
tude of the heterogeneity between studies from random-effect model, restricted maximum likelihood estimation (REML)
13
Journal of Neurology
heterogeneity between studies (I 2 = 87% [72–94%]; Although, the long-term follow-up (more than 5 years)
p < 0.0001). Pooled SARA progression rates differed appeared to be associated with faster disease progression in
between genotypes (p < 0.0001), with a faster progression all genotypes (figure S2), the difference was not statistically
in SCA1 compared to SCA6 (p < 0.0001), and in SCA3 com- significant (p for subgroup differences > 0.05). In SCA1, the
pared to SCA6 (p = 0.0052), while the disease progression pooled rates of annual progression were 1.77 (0.93–2.61;
was similar for the other comparisons (p > 0.0083 threshold p for heterogeneity equal to 0.05, I2 = 74.3%) for longer
with Bonferroni). According to the funnel plot and Egger’s follow-up against 1.73 (1.26–2.11, p = 0.71, I2 = 0.0%) for
test for asymmetry, except for patients with SCA1, no evi- shorter follow-up. The corresponding annual rates of pro-
dence for publication bias was found (figure S1). gression were 1.49 (1.36–1.62, p = 0.90, I2 = 0.0%) against
Based on the pooled effect sizes of the SARA progres- 1.21 (0.70–1.73, p = 0.11, I 2 = 56.1%) in SCA2, 1.57
sion, we calculated the sample sizes needed for two-group (1.45–1.69, p = 0.80, I2 = 0.0%) against 1.18 (0.15–2.21,
interventional trials of 1-year duration to achieve 80% and p < 0.0001, I 2 = 91.1%) in SCA3, and 0.83 (0.66–0.99,
90% power for each genotype with effects sizes ranging p = 0.44, I 2 = 0.0%) against 0.73 (0.29–1.18, p = 0.57,
from 20 to 100% (Fig. 2). With a power of 80%, 184 (92 I2 = 0.0%) in SCA6.
per group) patients with SCA1 would be needed to detect According to the geographical location of the studies,
a 50% reduction of pooled progression of the SARA score. the pooled rates of disease progression varied from one
The corresponding estimated number of patients for SCA2, SCA to another (figure S3). The only significant differ-
SCA3, and SCA6 were 194 (97 per group), 230 (115 per ence was observed for patients with SCA3 (p for interac-
group), and 860 (430 per group) respectively (Fig. 3). tion test = 0.0008). For SCA3 patients, disease progression
Because of the missing individual data, we were not able was faster in the Asian study with 1.60 (1.33–1.87, p = 0.51,
to identify independent factors influencing disease progres- I2 = 0.0%) and in the European study with 1.58 (1.45–1.70),
sion. Nevertheless, we realized a subgroup analysis to quan- and lower in American study with 0.65 (0.18–1.12). In
tify the impact of geographical location and study follow-up SCA1, the rate of disease progression was 2.07 (1.85–2.29,
on disease progression (Appendix). p = 0.34, I2 = 0.0%) in the European study, 1.61 (0.81–2.41)
Fig. 3 Sample size estimates. Required sample size per group in two-group interventional trials of 1-year duration for various effect sizes in
SCA1, SCA2, SCA3, and SCA6. SCA spinocerebellar ataxia
13
Journal of Neurology
in the American study, and 1.23 (0.39–2.07) in the Asian baseline score, the annual progression rate was 12% in the
study. The corresponding rates of disease progression Brazilian study compared to 11% noted in the European and
for patients with SCA2 were 1.47 (1.34–1.60, p = 0.37, 13% in the Asian study [18].
I2 = 0.0%), 1.52 (1.09–1.95), and 1.19 (0.18–2.21, p = 0.05, Identifying factors that influence disease course is crucial
I 2 = 75.0%), respectively, in the European, Asian, and to stratify subgroups in further clinical trials. The results
American study. For patients with SCA6, the corresponding from literature data about factors associated with disease
rates of disease progression were 0.78 (0.61–0.95, p = 0.33, progression were poorly understood and controversial.
I2 = 0.0%), 0.87 (0.32–1.42), and 1.00 (0.55–1.44, p = 0.81, Studies conducted in Europe identified shorter duration of
I2 = 0.0%), respectively, in the European, American, and follow-up, older age at inclusion, longer repeat expansions
Asian study. [7] and absence of pyramidal sign [4] as independent fac-
tors associated with faster disease progression for patients
with SCA1. While no factor has been identified in the others
Discussion parts of the world. For patients with SCA2, the correspond-
ing factors from European studies were short duration of
This first systematic review and meta-analysis of the natu- follow-up, lower age at disease onset [2, 4, 7], lower age at
ral history of the most common spinocerebellar ataxias inclusion, male sex, presence of the posterior column dys-
provides quantitative data on the progression of the most function, lower normal repeat CAG, and lower SARA score
common spinocerebellar ataxias based on patient numbers at baseline, while from Asian study [9], they were longer
that exceed those of previous studies and confirms that dis- repeat expansions, longer SARA score at baseline, and lower
ease progression was fastest in patients with SCA1, inter- disease duration. No factor was identified by the American
mediate in patients with SCA2 and SCA3, and slowest in (Brazilian and US) studies [5, 8]. For patients with SCA3,
patients with SCA6. Differences in these progressions could longer repeat expansions, longer SARA scores, and longer
be explained by the genetic background. In addition, our disease duration were associated with faster disease pro-
meta-analysis allows the approximated calculation of dis- gression in Chinese patients [9] while no factor has been
ease-specific sample sizes required for interventional trials identified in the remaining populations. For patients with
of SCAs. Sample sizes ranging from 184 to 230 patients SCA6, only the lower SARA score was associated with a
with SCA1, SCA2, and SCA3 would be needed to demon- change of disease progression from European and Chinese
strate the 50% reduction of progression of the SARA score populations [7, 9].
in a 1-year interventional trial. This is in line with previous As a measure of the disease progression of ataxia, we
reports by Jacobi (142–202) [7] and Lin (148–191) [9] from used the SARA score. SARA provides a semi-quantitative
different international cohorts and underlines the assumption assessment of ataxia on an impairment level. SARA was
that international clinical trials in this rare disease seem fea- validated in rigorous process [10] and is widely used in the
sible. Nevertheless, our sample size calculation is based on clinical assessments and clinical trials in ataxic disorders.
simulations to approximate the residual error of the pooled The limits of this study included the conducted post hoc
effect size, and individual-patients’ data meta-analyses are meta-analyses, based on available data with various sources
need to provide an exact sample from different populations. of heterogeneity (e.g., study population, length of follow-up
Subgroup analysis shows that the rates of disease progres- and statistical methods). Second, the missing reported esti-
sion were comparable according to the study follow-up dura- mates on factors influencing disease progression, negating
tion and geographical location of the studies except for patients to evaluate the impact of these factors on disease course.
with SCA3, in whom disease progression was faster in the Third, although the minimum number of studies required to
Asian and European than in the US studies. The lack of power assess the risk of publication bias through the Egger test is
could explain the absence of these differences. Individual- 10, we assessed this risk for each genotype. The absence of
patient data meta-analysis is needed to better understand the evidence of publication bias found except in patients with
risk factors that influence disease progression and improve SCA1 should, therefore, be interpreted with caution. Never-
patient stratification in interventional trials. There are also two theless, the quality of studies evaluated by the NOQAS being
Brazilian studies accessing the disease progression in SCA3, comparable, this risk of publication bias seems minimal.
one using the International Cooperative Ataxia Rating Scale
(ICARS) [16] and one the Neurological Examination Score
for the assessment of Spinocerebellar Ataxia (NESSCA scale) Conclusion and future direction
[17]. As NESSCA assesses both, ataxia and non-ataxia symp-
toms, a direct comparison to SARA is not possible. The second This meta-analysis extends our knowledge of the biological
of the Brazilian studies used the ICARS, which is a pure ataxia characteristics of patients with SCA1, SCA2, SCA3, and
scale. When relating the change over the observed time to the SCA6. Our findings have substantial implications on future
13
Journal of Neurology
clinical trials in SCAs, they provide information of annual 5. Ashizawa T, Figueroa KP, Perlman SL, Gomez CM, Wilmot GR,
disease progression from different populations as well as Schmahmann JD et al (2013) Clinical characteristics of patients
with spinocerebellar ataxias 1, 2, 3 and 6 in the US; a prospective
sample size estimates to detect a 50% reduction in disease observational study. Orphanet J Rare Dis 8:177
progression for each genotype. Nevertheless, individual- 6. Yasui K, Yabe I, Yoshida K, Kanai K, Arai K, Ito M et al (2014)
patient data meta-analysis are needed to better understand A 3-year cohort study of the natural history of spinocerebellar
the risk factors that influence disease progression. ataxia type 6 in Japan. Orphanet J Rare Dis 9(1):118
7. Jacobi H, du Montcel ST, Bauer P, Giunti P, Cook A, Labrum
R et al (2015) Long-term disease progression in spinocerebellar
Author contributions All authors contributed to the study conception ataxia types 1, 2, 3, and 6: a longitudinal cohort study. Lancet
and design. AD and HJ had the idea for the article and performed the Neurol 14(11):1101–1108
literature search, AD performed the statistical analysis, AD and HJ 8. On behalf of Rede Neurogenética, Monte TL, da Reckziegel ER,
drafted the article, and ST and TK critically revised the work. Augustin MC, Locks-Coelho LD, Santos ASP et al (2018) The
progression rate of spinocerebellar ataxia type 2 changes with
Funding None. stage of disease. Orphanet J Rare Dis Internet 13(1):20
9. Lin Y-C, Lee Y-C, Hsu T-Y, Liao Y-C, Soong B-W (2019) Compa-
Compliance with ethical standards rable progression of spinocerebellar ataxias between Caucasians
and Chinese. Parkinsonism Relat Disord 62:156–162
10. Schmitz-Hubsch T, du Montcel ST, Baliko L, Berciano J, Boesch
Conflicts of interest TK received research support from the Deutsche S, Depondt C et al (2006) Scale for the assessment and rat-
Forschungsgemeinschaft (DFG), the Bundesministerium für Bildung ing of ataxia: development of a new clinical scale. Neurology
und Forschung (BMBF), the Bundesministerium für Gesundheit 66(11):1717–1720
(BMG), the Robert Bosch Foundation, the European Union (EU), and 11. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD,
the National Institutes of Health (NIH). He has received consulting Rennie D et al (2000) Meta-analysis of observational studies
fees from Biohaven and UBC. He has received a speaker honorarium in epidemiology: a proposal for reporting. Meta-analysis Of
from Novartis. HJ received funding from the Medical Faculty of Hei- Observational Studies in Epidemiology (MOOSE) group. JAMA
delberg. AD, HJ, TK, and STDM declare no conflicts of interest. 283(15):2008–2012
12. Moher D, Liberati A, Tetzlaff J, Altman DG, The PRISMA Group
Ethical standards We confirm that the approval of an institutional (2009) Preferred reporting items for systematic reviews and meta-
review board was not required for this work and we have read the analyses: the PRISMA statement. PLoS Med 6(7):e1000097
Journal’s position on issues involved in ethical publication and affirm 13. Stang A (2010) Critical evaluation of the Newcastle-Ottawa scale
that this work is consistent with those guidelines. for the assessment of the quality of nonrandomized studies in
meta-analyses. Eur J Epidemiol 25(9):603–605
14. Egger M, Smith GD, Schneider M, Minder C (1997) Bias
in meta-analysis detected by a simple, graphical test. BMJ
References 315(7109):629–634
15. Higgins JPT, Thompson SG (2002) Quantifying heterogeneity in
1. Klockgether T, Mariotti C, Paulson HL (2019) Spinocerebellar a meta-analysis. Stat Med 21(11):1539–1558
ataxia. Nat Rev Dis Primer 5(1):24 16. Trouillas P, Takayanagi T, Hallett M, Currier RD, Subramony SH,
2. Klockgether T, Lüdtke R, Kramer B, Abele M, Bürk K, Schöls Wessel K et al (1997) International Cooperative Ataxia Rating
L et al (1998) The natural history of degenerative ataxia: a retro- Scale for pharmacological assessment of the cerebellar syndrome.
spective study in 466 patients. Brain J Neurol 121(4):589–600 J Neurol Sci 145(2):205–211
3. Lee Y, Liao Y, Wang P, Lee I-H, Lin K, Soong B (2011) Com- 17. Jardim LB, Hauser L, Kieling C, Saute JAM, Xavier R, Rieder
parison of cerebellar ataxias: a three-year prospective longitudinal CRM et al (2010) Progression rate of neurological deficits in a
assessment. Mov Disord 26(11):2081–2087 10-year cohort of SCA3 patients. Cerebellum 9(3):419–428
4. Tezenas du Montcel S, Charles P, Goizet C, Marelli C, Ribaï P, 18. França MC, D’Abreu A, Nucci A, Cendes F, Lopes-Cendes I
Vincitorio CM, Anheim M, Guyant-Maréchal L, Le Bayon A, (2009) Progression of ataxia in patients with Machado–Joseph
Vandenberghe N, Tchikviladzé M, Devos D, Le Ber I, N’Guyen disease. Mov Disord 24(9):1387–1390
K, Cazeneuve C, Tallaksen C, Stevanin G, Brice A, Durr A (2012)
Factors influencing disease progression in autosomal dominant
cerebellar ataxia and spastic paraplegia. Arch Neurol 69:500
13