 |
|
Friedreich Ataxia
Effects of Genetic Understanding on Clinical Evaluation and Therapy
David R. Lynch, MD, PhD;
Jennifer M. Farmer, MS;
Laura J. Balcer, MD, MSCE;
Robert B. Wilson, MD, PhD
Arch Neurol. 2002;59:743-747.
ABSTRACT
The discovery of the genetic cause of Friedreich ataxia has significantly
affected our understanding of the disorder at both the clinical and basic
science levels. Friedreich ataxia results from a deficiency of functional
frataxin, a protein that appears to be involved in mitochondrial iron homeostasis.
This leads to iron accumulation and mitochondrial abnormalities with consequent
oxidant damage. The clinical spectrum of Friedreich ataxia has also expanded
with the recognition of broader phenotypic features, including the absence
of classical Friedreich ataxia features, later age at onset, and spasticity
instead of ataxia. Although no proven therapy is yet available, antioxidants
are a potential method for therapeutic intervention.
INTRODUCTION
Friedreich ataxia (FRDA) is a progressive neurodegenerative disorder
with a prevalence of about 1 in 50 000 individuals, making it the most
common early-onset hereditary ataxia. Although sometimes viewed as a form
of cerebellar disease, the major pathological abnormalities are located outside
the cerebellum.1 Degeneration of the large
fiber dorsal root ganglion neurons and their axons in the dorsal columns gives
rise to loss of proprioception and a severe sensory ataxia. The spinocerebellar
tracts are also affected, along with a few other nuclei (including the dentate
nucleus of the cerebellum) within the central nervous system. Other clinical
manifestations include cardiomyopathy, scoliosis, and diabetes mellitus. In
recent years, the discovery of the disease-related gene, its mutation, and
its gene product have provided a genetic basis for redefinition of the illness
and new insight into the pathophysiological mechanisms of this disease.2 This review will describe how these discoveries have
affected the clinical diagnosis and evaluation of FRDA and discuss the proposed
pathophysiological mechanisms and potential therapies of this disease.
GENETIC BASIS OF FRDA
The major genetic mutation of FRDA is the presence of an expanded GAA
trinucleotide repeat in the first intron of both alleles of a novel gene (X25 or FRDA) encoding the protein
"frataxin." Normal alleles have fewer than 33 consecutive GAA repeats in the
gene, whereas abnormal alleles carry 66 to 1500 GAA repeats.2
Although 95% of people with FRDA have this abnormality, about 5% carry a point
mutation in the gene on one allele in association with an expanded repeat
on the opposite allele.2-6
This intronic trinucleotide repeat leads to the disruption of transcription
and the almost complete absence of frataxin. The length of the smaller repeat
correlates with the severity of the disease phenotype as assessed by age at
onset of the disease, suggesting that in shorter (but still pathologic) repeats,
a small amount of residual frataxin synthesis occurs (reviewed by Patel and
Isaya7). The disease-causing point mutations
also decrease frataxin levels, presumably by creating abnormalities of protein
stability, targeting, or function. Thus, the basic genetic mechanism of FRDA
is felt to be a decrease in functional frataxin.
PATHOPHYSIOLOGICAL UNDERSTANDING FROM THE GENE DISCOVERY
Much of what we have learned about frataxin comes from studies of the
yeast frataxin homologue, YFH1p. Both frataxin and YFH1p localize to mitochondria
at the matrix side of the inner mitochondrial membrane. Frataxin can substitute
for YFH1p in yeast, indicating that the 2 proteins are functional, as well
as structural, homologues. In yeast, loss of frataxin leads to mitochondrial
iron accumulation (reviewed by Patel and Isaya7).
The mitochondrial iron overload is associated with cytoplasmic iron depletion,
induction of plasma-membrane iron uptake proteins, decreased activities of
iron-sulfur cluster enzymes, sensitivity to oxidants, loss of mitochondrial
DNA, and impaired respiration. However, the exact role of frataxin in mitochondrial
dysfunction is not yet clear. The yeast protein YFH1p may act as an iron ligand
in the mitochondrial matrix. Isolated YFH1p binds Fe2+ (iron) and
keeps it in a reduced and available form, protects other biomolecules from
iron-induced oxidant damage, and can provide Fe2+ for heme synthesis
in vitro. Thus, the function of YFH1p (and, by extension, frataxin) may be
to bind Fe2+and decrease the rate of oxidation to Fe3+.
This would promote mitochondrial iron export, or use for iron-sulfur cluster
or heme synthesis, and prevent iron-induced free radical formation.
DEVELOPMENT OF MODEL SYSTEMS: CELLULAR AND ANIMAL
Confirmation of the role of frataxin in mitochondrial function requires
extension of the biochemical findings to higher animals such as mice, replication
in tissue culture paradigms, and demonstration of similar findings in tissue
from patients with FRDA. Iron deposits have been noted in the hearts of patients
with FRDA, and more recent experiments have demonstrated signs of oxidant
damage, decreased activities of iron-sulfur cluster enzymes, and loss of mitochondrial
DNA in cardiac tissue from patients with FRDA.7
Thus, the basic biochemical features derived from in vitro studies are present
in human tissue from biopsy or autopsy. Genetically engineered whole-animal
models also appear to replicate some features of FRDA. A homozygous null (knockout)
mouse lacking frataxin has been made, but these animals die before birth,
indicating that at least some frataxin function is necessary for survival.8 Tissue-specific knockout mice develop phenotypic changes
that overlap with FRDA, including hypertrophic cardiomyopathy and selective
degeneration of the large fiber sensory afferents; however, they also develop
changes (spongiform cerebral cortical degeneration) not characteristic of
FRDA.9 Overall, their phenotype is somewhat
more severe than that of FRDA.
Primary cell cultures provide an alternative model for FRDA. Cultured
fibroblasts and lymphocytes from patients with FRDA accumulate mitochondrial
iron and are sensitive to oxidant stress.7
Therefore, cell culture models correspond well with pathological findings
from patients and with mouse models and may be useful for screening compounds
for possible therapeutic agents.
EFFECT OF GENOTYPE ON DIAGNOSTIC POSSIBILITIES
The clinical phenotype of FRDA was well defined before genetic testing,
based on the examination of a large number of individuals.1
The clinical definition of FRDA includes autosomal recessive inheritance,
onset before age 25 years (and usually much earlier), progressive gait and
limb ataxia, absent lower extremity reflexes, and extensor plantar responses.
Electrophysiologically, FRDA is associated with absent or markedly reduced
sensory nerve action potentials with normal motor nerve conduction velocities,
and imaging findings of the brain are usually normal early in the disease.10 Cervical spine magnetic resonance imaging frequently
reveals atrophy. Most patients also have dysarthria, complete areflexia, pyramidal
weakness of the legs, loss of vibration and proprioception sensation, scoliosis,
and an electrocardiogram with abnormal results. Less common features include
nystagmus, optic atrophy, sensorineural hearing loss, distal motor atrophy,
pes cavus, and diabetes mellitus.
The clinical spectrum of FRDA has expanded noticeably through the use
of molecular diagnosis, now available in some form at several commercial laboratories.
Many patients with atypical features for FRDA (such as severe optic atrophy)
have genetically confirmed disease.4 Conversely,
the classic features of FRDA may not be present in patients with genetically
confirmed disease.11 Patients with FRDA may
retain reflexes and may present at substantially later ages than defined previously.12-13 This has led to the recognition of
a number of variant presentations, including diffuse chorea, spastic paraparesis,
and sensory neuronopathy.14-15
Patients with these atypical or variant features are more likely to
carry slightly different genotypes as well. Such individuals are more likely
to be among the 2% to 5% of patients harboring a point mutation on one allele.
In general, C terminal point mutations cause patients to be severely affected
and have more classical phenotypes, whereas N terminal point mutations result
in less severe abnormalities.6 However, there
are exceptions to this generalization.4 Alternatively,
patients with these atypical features (particularly spastic paraparesis) may
carry expanded GAA repeats that are longer than the normal range but not into
the range associated with childhood onset.12, 15
Patients with atypical clinical features for FRDA but with appropriate family
histories thus deserve testing for FRDA in some situations (Figure 1). However, in such cases, the laboratory performing the
testing must be capable of testing for point mutations to confirm the diagnosis.
The carrier frequency in the population for FRDA is about 1:100 (based on
an incidence of 1:40 000). Thus, some individuals with other neurodegenerative
disorders unrelated to frataxin will coincidentally appear to be carriers
for FRDA when assessed for the expanded GAA repeat of FRDA. This is the same
result seen in patients with FRDA who carry point mutations and a single expanded
allele. Confirmation of a variant type of FRDA in this situation requires
identification of a point mutation on the opposite allele.
|
|
|
|
Conceptual paradigm for use of Friedreich ataxia (FRDA) diagnostic
testing when evaluating a patient with an unknown ataxia. The ideal method
for evaluation will vary based on the availability of genetic testing and
other diagnostic testing for any individual patient. ARSACS indicates autosomal
recessive spastic ataxia of Charlevoix-Saguenay.
|
|
|
Although the genotypic changes are helpful for confirmation of disease,
they do not provide complete explanations for differences in disease phenotype.
The presence of cardiomyopathy is associated with longer GAA repeat length,
but not all patients with long GAA repeat lengths develop cardiomyopathy.14 In FRDA, the length of the shorter GAA repeat correlates
with age at onset,11 but the correlation is
not high enough (0.7-0.8) to predict reliably the age at onset or phenotype
in an individual patient. Substantial differences in age at onset have been
noted within a single generation of families and between families.16 This is illustrated by the Acadian variant of FRDA.17 Patients with this variant progress more slowly and
are, overall, less severely affected than patients with the classical phenotype.
However, these individuals share the same expanded GAA repeat found in others
with classical FRDA, and, as yet, no differences within the FRDA gene explain
the variation. This suggests that influences outside the FRDA gene may lead
to the milder clinical course.
A second group of patients has also been identified through genetic
testing: those who appear to have FRDA clinically but who carry no expanded
alleles.18-19 Theoretically, testing
for the expanded GAA repeat may miss the 0.05% of people with FRDA who harbor
2 point mutations. No patient with 2 point mutations has yet been described.
In addition, several families have been characterized with the phenotype of
FRDA whose disease is linked to a different genetic locus (FRDA2).20 More likely, the individuals
who test entirely negative for FRDA expansions have other less common diseases.
A variety of disorders clinically overlap with FRDA, either in the presence
of a primarily sensory ataxia or in the age at onset. There are many defined
diagnostic possibilities for these individuals, a few of which deserve special
mention due to recent developments:
- Ataxia with vitamin E deficiency: This inherited disorder was initially described both genetically as
a subgroup of patients with FRDA linked to a separate genetic locus and by
the discovery of an FRDA phenotype associated with low levels of vitamin E.
The phenotype of ataxia with vitamin E deficiency is essentially identical
to that of FRDA, with the exception of retinopathy in a few patients.21 More important, it represents the major treatable
disorder overlapping with FRDA. Patients with ataxia with vitamin E deficiency
stabilize and may improve following vitamin E treatment, which must be lifelong.
- Autosomal recessive spastic ataxia
of Charlevoix-Saguenay (ARSACS): This genetically distinct disorder
is found in individuals from a specific region of Quebec, Canada, thus overlapping
with the population at risk for FRDA. However, as the name implies, spasticity
is seen in this disorder, distinguishing it from classical FRDA, and cerebellar
atrophy is found pathologically and on imaging studies. Genetically, the disease
is defined by mutations in a novel protein on chromosome 13.22
A similar disorder inherited in an autosomal recessive pattern in Tunisians
has been mapped to the same region of chromosome 13.23
- Posterior column ataxia with
retinal pigmentary changes: This unique disorder, which is localized
to chromosome 1q31-32, is distinguished from FRDA by the retinal changes.24
- Early-onset cerebellar atrophy
with retained reflexes25: This disorder
likely constitutes a heterogeneous syndrome including individuals who overlap
clinically with all of the above groups and atypical FRDA but in whom no definitive
genetic or biochemical diagnosis can be confirmed.
On occasion, individuals with Charcot-Marie-Tooth disease or mitochondrial
disorders may resemble patients with FRDA. The signs and symptoms of FRDA
are reminiscent of the mitochondrial encephalomyopathies with peripheral neuropathy,
ataxia, hearing loss, diabetes mellitus, and cardiac disease. In addition,
rare patients with FRDA develop paroxysmal features characteristic of mitochondrial
disorders.26 However, in spite of their pathophysiological
similarity, FRDA and mitochondrial disorders are usually readily distinguishable
on clinical grounds.
GENETIC COUNSELING
The availability of genetic testing for FRDA has clarified some issues
in genetic counseling of patients with ataxia, while introducing challenges
in other ways. The standard genetic testing offered by commercial laboratories
measures the GAA repeat length on both alleles. This testing offers a 95%
to 98% detection rate, which makes it useful for confirming diagnoses and
carrier testing. Patients with FRDA can constitute a significant proportion
of patients with apparently sporadic ataxia due to autosomal recessive inheritance,
even in the adult-onset population.18 Genetic
testing allows these individuals to be given a firm diagnosis. Genetic testing
also identifies carriers of FRDA among individuals at increased risk based
on family history and among spouses of carriers or affected individuals. The
general population carrier frequency is roughly 1 in 100 persons. As with
many trinucleotide repeat disorders, genetic counseling for individuals who
are carriers for repeat lengths at the shorter end of the pathologic range
can be more complex. In addition, the relatively small number of independent
patients with point mutations makes prediction of the exact phenotype in these
cases very difficult. Although carrier testing can improve overall risk assessment,
it cannot always define the phenotype.
THERAPY: PRESENT AND FUTURE
At present there is no clearly effective protective pharmacological
therapy for FRDA. Citalopram hydrobromide, amantadine hydrochloride, and serotonin
derivatives have shown minimal to no benefit as symptomatic therapy in small
trials.27-29 However,
comprehensive care of the patient with FRDA is not passive, particularly with
the potential for more novel therapies in the tangible future. Aggressive
surveillance for progressive cardiomyopathy, arrhythmias, scoliosis, and diabetes
mellitus can markedly improve the life span and quality of life of the patient.
Such care is also crucial for maintaining function for future therapeutic
approaches.
The overlapping phenotypes of vitamin E deficiency and FRDA and the
biochemical studies on frataxin have led to investigations designed to develop
antioxidant therapy in FRDA. Noninvasive measures of patients support the
link to antioxidants as a potential therapy. 31Phosphorous 31–labeled
magnetic resonance spectroscopy has demonstrated that mitochondrial adenosine
triphosphate production is decreased in the skeletal muscles of patients with
FRDA.30-31 Urinary measurement
of 8-hydroxy-2'-deoxyguanosine and plasma malondialdehyde, markers of
reactive oxygen species production, are elevated in patients with FRDA and
may be useful as surrogate markers of pathophysiological characteristics in
patients with FRDA.32-33
These findings have expanded antioxidant therapy in FRDA. Using markers
of disease activity (such as cardiac ejection fraction) as outcome measures,
FRDA has been treated with some success in an unblinded fashion with a coenzyme
Q analogue (idebenone), but a more recent study using a shorter duration of
therapy did not reveal any beneficial effect.34-35
The successful study demonstrated a decrease in cardiac hypertrophy, suggesting
a capacity for repair of damaged heart muscle cells. Other cardiac surrogate
markers improve with high doses of the antioxidant coenzyme Q and vitamin
E.31 Further trials will be needed to define
the specific features of disease that respond to antioxidants, as well as
the appropriate doses and lengths of therapies.
However, many difficulties may be present in the performance of such
trials. With the availability of many antioxidants on a nonprescription basis,
the widespread use of these well-tolerated dietary agents may complicate a
placebo-controlled trial. Although the ease of administering such antioxidants
suggests little benefit to a systematic examination of their usefulness in
FRDA, other factors demonstrate the need for such an analysis. Antioxidants
are readily available, but they are not necessarily inexpensive. In addition,
although no adverse effects are commonly reported in moderate duration use,
the long-term effects are not known. Finally, it is possible that use of multiple
antioxidants is counterproductive. In one in vitro model of FRDA, coenzyme
Q improved iron-sulfur–containing enzyme activities, vitamin E had no
effect, and ascorbate worsened enzymatic activities.36
These results demonstrate the need for a systematic understanding of the usefulness
of dietary antioxidants in disorders such as FRDA.
CLINICAL OUTCOME MEASURES IN FRDA
Although initial therapeutic trials are now being conducted for FRDA
largely using surrogate markers of disease activity, more systematic trials
will require the use of clinical outcome measures. Successful use of agents,
particularly those with significant adverse effects, requires a demonstration
of effects on patient outcome, not simply an improvement in surrogate markers.
Although magnetic resonance spectroscopy measures correlate with triplet repeat
length, they do not clearly correlate with disease progression.30, 37
In addition, the expense and technical expertise needed for performance of 31P–magnetic resonance spectroscopy may make it
impractical for large-scale clinical trials. Echocardiography detects disease
effects on the heart but provides no information on neurological features.
At present, no adequate measure has been identified or validated in a large
cohort of patients with FRDA. Clinical outcome measures must be easy to administer
and sensitive to change over time. The ideal scale for FRDA would be sensitive
and yet capture the entire clinical neurological spectrum of the disease.
The most sensitive clinical outcome measures for multiple sclerosis now reflect
simple performance measures rather than traditional neurological examinations
or disability scales. Although the pathophysiological characteristics of multiple
sclerosis are distinct from those of FRDA, both lead to progressive dysfunction
of multiple neurological systems with variable phenotypes. Future measures
of FRDA might use a similar design to that of the Multiple Sclerosis Functional
Composite,38 and development of analogous measures
must represent an avenue for clinical research in FRDA if therapeutic trials
are eventually to become effective.
AUTHOR INFORMATION
Accepted for publication October 11, 2001.
Author contributions: Study concept and design (Drs Lynch, Balcer, and Wilson and Ms Farmer); drafting
of the manuscript (Drs Lynch and Balcer and Ms Farmer); critical revision of the manuscript for important intellectual content (Dr Wilson); administrative, technical, and material support (Dr Lynch and Ms Farmer); study supervision (Drs Lynch and Balcer and Ms Farmer).
This study was supported by grants NS39126 (Dr Lynch) and EY00351 (Dr
Balcer) from the National Institutes of Health, Bethesda, Md; a Beeson Scholar
Award from the American Federation for Aging Research, New York, NY (Dr Lynch);
and a grant from the National Ataxia Foundation, Minneapolis, Minn (Dr Wilson).
Corresponding author and reprints: David R. Lynch, MD, PhD, Division
of Neurology, Children's Hospital of Philadelphia, 502 Abramson Bldg, Philadelphia,
PA 19104-44318 (e-mail: lynch{at}pharm.med.upenn.edu).
From the Departments of Neurology (Drs Lynch and Balcer), Pediatrics
(Dr Lynch), Medicine (Ms Farmer), Ophthalmology (Dr Balcer), and Pathology
and Lab Medicine (Dr Wilson), University of Pennsylvania School of Medicine,
Philadelphia.
REFERENCES
| |
1. Harding AE. Friedreich's ataxia: a clinical and genetic study of 90 families with
an analysis of early diagnostic criteria and intrafamilial clustering of clinical
features. Brain. 1981;104:589-620.
FREE FULL TEXT
2. Campuzano V, Montermini L, Molto MD, et al. Friedreich's ataxia: autosomal recessive disease caused by an intronic
GAA triplet repeat expansion. Science. 1996;271:1423-1427.
ABSTRACT
3. Bidichandani SI, Ashizawa T, Patel PI. Atypical Friedreich ataxia caused by compound heterozygosity for a
novel missense mutation and the GAA triplet-repeat expansion. Am J Hum Genet. 1997;60:1251-1256.
ISI
| PUBMED
4. McCormack ML, Guttmann RP, Schumann M, et al. Frataxin point mutations in 2 patients with Friedreich's ataxia and
unusual clinical features. J Neurol Neurosurg Psychiatry. 2000;68:661-664.
FREE FULL TEXT
5. Forrest SM, Knight M, Delatycki MB, et al. The correlation of clinical phenotype in Friedreich ataxia with the
site of point mutations in the FRDA gene. Neurogenetics. 1998;1:253-257.
FULL TEXT
|
ISI
| PUBMED
6. Cossee M, Durr A, Schmitt M, et al. Friedreich's ataxia: point mutations and clinical presentation of compound
heterozygotes. Ann Neurol. 1999;45:200-206.
FULL TEXT
|
ISI
| PUBMED
7. Patel PI, Isaya G. Friedreich ataxia: from GAA triplet-repeat expansion to frataxin deficiency. Am J Hum Genet. 2001;69:15-24.
FULL TEXT
|
ISI
| PUBMED
8. Cossee M, Puccio H, Gansmuller A, et al. Inactivation of the Friedreich ataxia mouse gene leads to early embryonic
lethality without iron accumulation. Hum Mol Genet. 2000;9:1219-1226.
FREE FULL TEXT
9. Puccio H, Simon D, Cossee M, et al. Mouse models for Friedreich ataxia exhibit cardiomyopathy, sensory
nerve defect and Fe-S enzyme deficiency followed by intramitochondrial iron
deposits. Nat Genet. 2001;27:181-186.
FULL TEXT
|
ISI
| PUBMED
10. Ormerod IE, Harding AE, Miller DH, et al. Magnetic resonance imaging in degenerative ataxic disorders. J Neurol Neurosurg Psychiatry. 1994;57:51-57.
ABSTRACT
11. Durr A, Cossee M, Agid Y, et al. Clinical and genetic abnormalities in patients with Friedreich's ataxia. N Engl J Med. 1996;335:1169-1175.
FREE FULL TEXT
12. Lhatoo SD, Rao DG, Kane NM, Ormerod IE. Very late onset Friedreich's presenting as spastic tetraparesis without
ataxia or neuropathy. Neurology. 2001;56:1776-1777.
FREE FULL TEXT
13. Klopstock T, Chahrokh-Zadeh S, Holinski-Feder E. Markedly different course of Friedreich's ataxia in sib pairs with
similar GAA repeat expansions in the frataxin gene. Acta Neuropathol (Berl). 1999;97:139-142.
FULL TEXT
| PUBMED
14. Pandolfo M. Molecular pathogenesis of Friedreich ataxia. Arch Neurol. 1999;56:1201-1208.
FREE FULL TEXT
15. Ragno M, De Michele G, Cavalcanti F, et al. Broadened Friedreich's ataxia phenotype after gene cloning: minimal
GAA expansion causes late-onset spastic ataxia. Neurology. 1997;49:1617-1620.
FREE FULL TEXT
16. Bidichandani SI, Garcia CA, Patel PI, Dimachkie MM. Very late-onset Friedreich ataxia despite large GAA triplet repeat
expansions. Arch Neurol. 2000;57:246-251.
FREE FULL TEXT
17. Montermini L, Richter A, Morgan K, et al. Phenotypic variability in Friedreich ataxia: role of the associated
GAA triplet repeat expansion. Ann Neurol. 1997;41:675-682.
FULL TEXT
|
ISI
| PUBMED
18. Geschwind DH, Perlman S, Grody WW, et al. Friedreich's ataxia GAA repeat expansion in patients with recessive
or sporadic ataxia. Neurology. 1997;49:1004-1009.
FREE FULL TEXT
19. McCabe DJ, Ryan F, Moore DP, et al. Typical Friedreich's ataxia without GAA expansions and GAA expansion
without typical Friedreich's ataxia. J Neurol. 2000;247:346-355.
FULL TEXT
|
ISI
| PUBMED
20. Christodoulou K, Deymeer F, Serdaroglu P, et al. Mapping of the second Friedreich's ataxia (FRDA2) locus to chromosome
9p23-p11: evidence for further locus heterogeneity. Neurogenetics. 2001;3:127-132.
FULL TEXT
|
ISI
| PUBMED
21. Cavalier L, Ouahchi K, Kayden HJ, et al. Ataxia with isolated vitamin E deficiency: heterogeneity of mutations
and phenotypic variability in a large number of families. Am J Hum Genet. 1998;62:301-310.
FULL TEXT
|
ISI
| PUBMED
22. Engert JC, Berube P, Mercier J, et al. ARSACS, a spastic ataxia common in northeastern Quebec, is caused by
mutations in a new gene encoding an 11.5-kb ORF. Nat Genet. 2000;24:120-125.
FULL TEXT
|
ISI
| PUBMED
23. Mrissa N, Belal S, Hamida CB, et al. Linkage to chromosome 13q11-12 of an autosomal recessive cerebellar
ataxia in a Tunisian family. Neurology. 2000;54:1408-1414.
FREE FULL TEXT
24. Higgins JJ, Morton DH, Loveless JM. Posterior column ataxia with retinitis pigmentosa (AXPC1) maps to chromosome
1q31-q32. Neurology. 1999;52:146-150.
FREE FULL TEXT
25. Klockgether T, Petersen D, Grodd W, Dichgans J. Early onset cerebellar ataxia with retained tendon reflexes: clinical,
electrophysiological, and MRI observations in comparison with Friedreich's
ataxia. Brain. 1991;114:1559-1573.
FREE FULL TEXT
|
