Spinocerebellar Ataxia 3 and Machado-Joseph

Disease: Clinical, Molecular, and

NeuropathoIogicaI Features
Alexandra Durr, MD,* Giovanni Stevanin, MS,* Gkraldine Cancel, MS," Charles Duyckaerts, MD, PhD,t
Nacer Abbas, PhD,* Olivier Didierjean, BS,* Her& Chneiweiss, MD, PhD,* Ali Benomar, MD,$
Olivier Lyon-Caen, MD,* Jean Julien, MD,$ Michel Serdaru, MD,* Christiane Penet, BS,*
Yves Agid, MD, PhD,* and Alexis Brice, MD*

Patients with spinocerebellar ataxia 3 (SCA3) and Machado-Joseph disease (MJD) carry an expanded CAG repeat in
the MJDl gene. One hundred twenty families of different geographic origin with autosomal dominant cerebellar ataxia
(ADCA)type I were tested. Thirty-four families (126 patients) carried an expanded CAG repeat. The expanded and the
normal allele did not overlap and the repeat was unstable during transmission, with variation in the size of the CAG
length ranging from -8 to +5 and a mean expansion of 0.86 repeats without differences according to the parental sex.
There was a combined effect of the number of CAG repeats of the expanded and normal allele on the age at onset,
which accounted for 70% of its variability. The length of the CAG repeat influenced the frequency of clinical signs
associated with cerebellar ataxia, such as abnormal tendon reflexes or decreased vibration sense, whereas the interindividual variation of supranuclear ophthalmoplegia, sphincter and swallowing difficulties, and amyotrophy was mostly determined by different disease durations. We compared the clinical profile of 91 SCA3IMJD patients with 51 SCAl and
32 SCA2 patients. There were striking differences between the SCA3/MJD and SCA2 but not with SCAl groups of
patients. Despite their clinical similarities, distinct neuropathological features were observed in 2 SCA3/MJD and 2
SCAl patients.
Durr A, Stevanin G, Cancel G, Duyckaerts C, Abbas N , Ilidierjean 0, Chneiweiss H, Benomar A,
Lyon-Caen 0, Julien J, Serdaru M, Penet C, Agid Y, Brice A. Spirocerebellar
ataxia 3 and Machado-Joseph disease: clinical, molecular, and
neuropathological features. Ann Neurol 1996;39:490-499

The autosomal dominant inherited cerebellar ataxias

(ADCAs), are phenotypically and genotypically heterogeneous. Three groups have been defined on the basis
of clinical distinctions, ADCA types I, 11, and I11 [ I ] .
Type I is genetically and clinically the most heterogeneous group. It is characterized by the variable association of cerebellar signs with supranuclear ophthalmoplegia, extrapyramidal signs, dementia, and
amyotrophy, and is caused by at least five genes [2].
SCAl , spinocerebellar ataxia 1, was mapped to chromosome 6 [3] and an expanded and unstable CAG
repeat was identified as the responsible mutation [4].
SCAl, with a frequency that varies from 14 [5] to 42%
[6],is not the major locus for ADCA iype I [7].
SCA2
was localized on chromosome 12 [8], then SCA3
and Machado-Joseph disease (MJD) on chromosome

14 [9, lo]. SCAS and MJD were both found to map

to the same 3 cM region in French and Brazilian-Portuguese families [ 11- 131, suggesting that the mutations were allelic. Identification of the gene MJDl
and its mutation, an abnormal expansion of CAG
repeats [14], showed that the same gene and unstable
mutation are involved in French and German SCA3
families [15, 161. Finally, SCA4 and SCAS were assigned to chromosomes 16 and 11 by linkage analysis
in single families, but, ro our knowledge, these localizations have not yet been confirmed in other kindreds
[17, 181.
In this study, we have determined the frequency of
the SCA3 and SCAl mutations in ADCA families and
have analyzed the associated clinical, molecular, and
neuropathological features.

From the *INSERM U289 and Fkdkration de Neurologie, and

?Service de Neuropathologie, H6pital de la Salpkrithc, Paris, and
$Service de Neurologie, HApital de Haut I.&\leque, Pessac, France;
and $Service de Neurologic HApital des Specialitks, Rabat, Morocco.

Received Aug 31, 1!)95, and in revised form Oct 30. Accepted for
publication Nov 6, 1395.

490

Address correspondrnce to Dr Diirr, INSERM u2s3, HApital de

75651 Paris Cedex 13,
la Salp;tritre, 47 boulevard de
France.

Copyright 0 1996 hy the American Neurological Association

Patients and Methods

Patients
One hundred seventy-three index patients were screened for
the SCA3IMJD mutation; 124 were French and 18 were
from other European countries, including Belgium (4), Italy
( 3 , Spain (3),Portugal (2), Poland (2), Germany (2), Serbia
(I), and Sweden (1). Sixteen families came from Morocco,
4 from Algeria, 3 from the French West Indies, 2 from Riunion Island and India, and 1 each from Ivory Coast,
Yemen, Brazil, and French Guyana. All were white, except
9 who were black; 4 were from Morocco, 2 from French
West Indies, and 1 each from Ivory Coast, French Guyana,
and Rkunion Island.
One hundred thirty-three index patients had ADCA.
Among those, 120 were classified as type I, 9 were type 11,
where the cerebellar ataxia was associated with retinal degeneration in all family members, and 4 were type 111, presenting
with pure cerebellar signs after a disease duration of more
than 10 years. We included in the type I group 1 family
(SAL-318) with the SCA3/MJD mutation and with neuropathological features suggestive of dentatorubropallidoluysian
atrophy (DRPLA) [16].
Thirty-one isolated cases with nonfamilial olivopontocerebellar atrophy, 7 patients with probable autosomal recessive
cerebellar ataxia with adult onset, 1 with a complex form of
autosomal dominant spastic paraplegia, which was associated
with cerebellar signs in all 4 affected members, and 1 with
epilepsy, cerebellar ataxia, and normal muscle biopsy, were
studied.
All consenting family members were examined using a
standardized procedure to establish age and first signs at onset, clinical profile, and disease duration. At least one thorough neurological examination was performed. Neuropathological examinations were carried out in 4 unrelated patients,
2 with the SCAl and 2 with the SCA3/MJD mutations.

given with the standard deviations. Pearsons R and regression coefficients were calculated to evaluace the correlation
between age at onset and length of CAG repeat.

Results

Churucterization of the SCBIMJD Mutution

An expanded CAG
repeat at the SCA31MJD locus was found in patients
from 34 of 120 families with ADCA type I, for 30
families for which detailed clinical information was
available. The SCA3/MJD mutation represented 28%
of all tested ADCA type I families, which is more than
twice the frequency of the SCAl mutation (13%) in
this population (Table 1). SCA3/MJD was found in
25 French families, 2 Moroccan families, and 1 each
from Belgium, Portugal, Spain, Algeria, French Guyana, Ivory Coast, and Yemen. Patients from all the
other families were either homozygous or heterozygous
for alleles with a number of CAG repeats in the normal
range (14-40).
The number of CAG repeats in the expanded allele
of 126 mutation carriers, 99 affected and 27 asymptomatic individuals, ranged from 64 to 82 with a median of 73 (Fig 1). The distribution of the expanded
alleles was unimodal, 60% of patients had CAG repeats
between 70 and 76. Normal chromosomes (n = 502)
from 193 unrelated unaffected controls and 116 normal chromosomes from patients and at-risk carriers
contained between 14 and 40 CAG repeats (see Fig
1). In the normal alleles, three predominant repeat sizes
were seen, ie, 14 (24%), 23 (32%),and 27 (12%). The
CAG repeat lengths in normal and expanded alleles did
not overlap.
FREQUENCYO F THE MUTATION.

Genotyping
Genotypes at the SCA3/MJD locus were determined as previously described [14, 161. The number of CAG repeats was
determined on gel electrophoresis in comparison with normal and expanded alleles that had been subcloned and sequenced.

Statistical Analysis
Comparisons of means were performed with nonparametric
tests and Students t test, and comparisons of frequencies
and Yates corrected
tests. Mean values are
with the

x2

x2

The mean
age at onset of 91 patients, for whom detailed clinical
information was known, was 36 t 12 years with a
unimodal distribution from 14 to 70 years. There was
a significant negative correlation between the size of
the CAG repeat and the age at onset ( r = -0.8, slope
= -2.66 years per CAG repeat, p < 0.001) (Fig 2).
There was no significant difference according to the
sex of the patient. Furthermore, the size of the CAG
CAG REPEAT LENGTH AND AGE AT ONSET.

Table 1. Frequency of the SCAl and SCA3 Mutations

Origin of the ADCA Type I Families

SCAl (Yo)
SCA3 (%)

French
(n = 87)

Other European
(n = 10)

North African
(n = 12)

Others
(n = 11)

Total
(n = 120)

17

20
30

8
17

9
27

13
28

30

SCA = spinocerebellar ataxia; ADCA = autosomal dominant cerebellar ataxia.

Diirr et al: SCA3 and MJD

491

35
30

25
n

20
t?
E
Q)

15
L

Lc

10

,.11111

LAW+--

14 16 18 20 22 24 26 28 30 32 34 36 38 40 42 44 60 62 64 66 68 70 72 74 76 78 SO 82

Number of CAG repeats

Fig 1. Distribution of CAG repeats on 1.26 pathological chromosomes from 34families with the spinocerebelhr ataxia 3/
Machado-Joseph disease mutation and on 502 normal chromosomes. Black bars = nomal chromosomes from patients; hatched
bars = controls.

repeat on the normal allele was also negatively correlated with the age at onset ( 7 = -0.25, slope = -0.57
years per CAG repeat, p < 0.05). This demonstrates
that the number of CAG repeats on the expanded and
the normal alleles account, respectively, for 64% and
6% of the variability in age at onset.

The mean age

at onset of offspring of affected fathers (n = 46) was
40 2 14 years and 34 2 9 years in offspring of affected
mothers (n = 43). The difference was statistically significant ( p = 0.045). The mean length of maternally
transmitted CAG repeats (73.8 2 3 ) was similar, however, to that of paternal transmissions (73.0 ? 4).
The CAG repeats were unstable during transmission.
The variation in repeat length ranged from - 8 to 5
with a mean expansion of +0.86 (ZAG repeats. In 22
maternal transmissions, the mean variation was 0.5
+_ 2.2 (median
1) and ranged from - 8 to +3 CAG
repeats. The mean expansion (+1.4 2 2; median,
+1.5; range, -2 to + 5 ) was larger in 14 paternal
transmitted alleles, but the difference was riot significant. The variance, observed in 22 sibships of maternal

70

CAG EXPANSION AND I I S TRANSMISSION

492 Annals of Neurology

Vol 39 No 4 April 1996

o/-60

65

70

75

80

85

Number of CAG repeat

Fig 2. Correhtion between age at onset (years) and CAG
repeat number.

(5.0) and in 14 sibships of paternal (4.0) transmissions

was also similar.
The age at onset and the number of CAG repeats
differed among generations. A mean increase in the
number of CAG repeats (GI-GIl, +3.67; GII-GIII,
1.49) was correlated with a younger age at onset in
successive generations ( p < 0.001). The oldest generation, GI (n = 9), had a mean age at onset of 53 ?

12 years, GI1 (n = 47) 41 t 10 years, and GI11

(n = 37) 29 2 7 years. O n e patient in GIV was already affected at age 21. Between GI and GII, in which
penetrance was almost complete, a mean anticipation
of 12 years was observed, in accordance with the expected value of 9.8 years, calculated from the regression
line for a mean increase of 3.7 CAG repeats. In contrast, in 13 parent-offspring pairs, for whom the age
at onset and the size of CAG expansion were known,
the mean anticipation was 15 years and the CAG difference 1.9. This was significantly more than the expected 5. I-year anticipation ( p < 0.00 1, paired t test),
indicating a marked observation bias in the assessment
of age at onset in parent-offspring pairs.
There was no correlation between the CAG repeat
length and the tendency to expansion. The mean
length of the CAG repeat was 71.9 in parents who
transmitted 1 CAG repeat or more, and 72.8 in parents who transmitted CAG repeats of the same length
or smaller. 'This difference was not significant and suggests that the larger size did not predispose to instability or transmission of larger CAG repeats. There was
no correlation between the age at conception and the
relative or absolute instability in either sex (data not
shown).
In sibships with at least 2 siblings, for which the
number of CAG repeats was determined in all siblings,
the ratio of carriers and noncarriers was the same in
paternal (42/31, 0.7) as in maternal (31/23, 0.7) offsprings. This indicates an absence of meiotic distortion.

Climical Annlysis of 30 SCH3IMJL) Families

with 91 Patients

ophthalnioplegia, swallowing and sphincter difficulties,

and amyotrophy, was significantly correlated with disease duration ( p < 0.01; Fig 3). Indeed, after dividing
patients into three groups with disease durations of less
than 5 years (n = 27), between 5 and 10 years (n =
3 9 , and more than 10 years (n = 26), ophthalmoplegia, amyotrophy, and swallowing and sphincter
difficulties were found to be significantly more frequent
in patients with longer disease duration ( p < 0.01).
This was not observed for reflexes in the lower limbs,
vibration sense, or abnormal somesthetic evoked potentials and axonal neuropathy.

m THF. SIZE 01: THE cnc

T h e frequency of clinical signs associated
with cerebellar ataxia differed significantly among patients with less than 71, between 71 and 74, and more
than 74 CAG repeats (Table 3). Mean age a t onset was
significantly different in these subgroups, but disease
duration was the same. Patients with more than 74
CAG repeats had frequently increased reflexes (56%)
but none of those with less than 71 CAG repeats. In
contrast, decreased reflexes were found in 54% of patients with the number of CAG repeats less than 71
(30% with Babinski sign) but only 3% of patients with
more than 74 CAG repeats (54% with Babinski sign).
Dysphagia, sphincter disturbances, supranuclear ophthalmoplegia, extrapyramidal signs, and bulging eyes
did not differ among the three groups. Axonal neuropachy was more frequent i n patients with less than 71
CAG (7 of 9) than in those with the largest repeats (4
of 10).

(:I.INICAL S I G N S I N RELATION

REPEAT.

FREQUENCIES O F CLINICAL SIGNS ASSOCIATED W I T H CER-

COMPARISON

patients had cerebellar ataxia,

which was associated with an extensor plantar response
in 46%, increased reflexes in 32%, decreased vibration
sense in 55%, and supranuclear ophthalmoplegia in
47% (Table 2). Dystonia and parkinsonian signs were
found in only 18% of the patients. Bulging eyes were
noticed in 23% of the patients. Funduscopy was normal in 49 patients, 3 had optic atrophy, 2 retinal degeneration, and 1 unilateral toxoplasmic lesions. Nineteen cerebral computed tomographic scans and 14
magnetic resonance imaging showed different degrees
of olivoporitocerebellar atrophy. Electrophysiological
studies of conduction velocities and electromyograms
(n = 30) revealed an axonal neuropathy in 60% of
the cases. T w o patients had signs of demyelination and
1 had pure lower motoneuron degeneration. Evoked
somesthetic potentials had abnormal latencies in 18 of
24 tested patients. Brainstem auditory evoked potential
latencies were increased in 13 of 25 patients, visual
evoked potentials in 4 of 27.
T h e frequency of signs associated with cerebellar
ataxia, dysarthria, and dysmetria, such as supranuclear

TIENTS.

EBELIAK ATAXIA.

OF

SCAl,

SCA2,

AND

SCA3/MJI>

PA-

T h e overall frequencies of signs and symptoms associated with the cerebellar syndrome were analyzed in 91 SCA3/MJD patients and compared with
5 1 SCAl ([7]; Durr A, unpublished results) and 31
SCA2 [I91 patients (see Table 2). Mean ages at onset
and at the time of medical examination, as well as disease durations, were not different among the groups.
Increased reflexes in the lower limbs were significantly
more frequent in SCA3 and SCAl than in SCA2.
Babinski sign was equally frequent in SCAl and
SCA3 patients but was less frequent in SCA2 subjects.
Memory impairment or dementia were found in
29% of SCA2 patients but were rare, 6% and 8Y0,
in SCA3 and SCA1, respectively. The frequency of
bulging eyes was similar in SCAl and SCA3 (27% and

23%).
Neiiropathology
T h e clinical characteristics of 4 patients with neuropathological examination are shown in Table 4. Although the ages at death were variable, disease durations until death were similar in the 2 SCA3/MJD

Durr et al: SCA3 and MJD 493

Tnblr 2. Clinical Cornpnrison OJ'SCAI, SCA2. mid SC&/IZ.IJL, Piltirnts

Patients (women / men)

Mean age a t onset (range) (yr)
Mean age at examination (yr)
Clinical signs associated with the
cerebellar syndrome (%)
Increased reflexes
Decreased or abolished reflexes
Extensor plantar reflexes
Amyot rophy
Decreased vibration sense
Supranuclear ophthalmoplegia
Slow saccades velocity alone
Dystonia
Extrapyramidal rigidity or bradykinesia or aniiniia
Swallowing difficulties
Sphincter disturbances
Menioiy i m pai rriien t o i- dementia

SCAl

SCA2"

SCA3/ MJD

5 1 (29/22)
34 ? 10 (21-55)
42 5 13

31 ( 1 5 / 1 6 )
31 5 17 (8-67)
42 5 18

91 (50141)

30
10"

6'
68
23''
26
32
52
13
3
6

32
28
46
12
55
47
16
9

36
42
29'

46
35
6

45
10

49
28
4
6
16

56
24
8

36 ? 12 (14-70)
46 5 14

"Ilara from reference 7 and A. Diirr (unpublished results)

hDara from refrrencc 19.

Statistically significant differences hetween SCA.l a n d SCA3 and SCA? and SCA3 were indicated by ' ( p < 0.01); "((p < 0.05); and ' ( p <
0.001).

SCA

spInocerebrl~araraxin; MJD = Machado-Joseph disease

Fig 3. F r e q ~ ~ e of
i q ~clinicnl sigm cmocirited iiiith cerebellar ntlixin according to diseilsr dzrration among groups. 'p < 0.01:
"*p < 0.001.

494 Annals of Neurology

Vol 33

No 4

April 1996

Table 3. Clinical Differences as a Functioii oftbe N~rmberof CAG Repeclts in SCA33/MjD Patients

Number of CAG Repeats

64 < n

Patients
Mean age at onset (yr)
Mean disease duration (yr)
Increased reflexes> (Yo)
Decreased or absent reflexes (Yo)
Decreased or abolished vibration sense (%)

< 71

71

24

54

32
38
10
28
31

87

65

43

% 11

10 -c 8
0

5 11 5

74

>74
32
27 -t 6
7 t 6
56
3
28

*7
*8

p < 0.001, among groups.

SCA = spinocerebellar ataxia; M J D = Machado-Joseph disease.

Table 4. Clinical Cliaracteristics of Patients u h b Neuropatliological Examination

SCA3/ MJD

Sex
Number of CAG repeats
Age at onset (yr)
Age at examination (yr)
Disease duration (yr)
Age at death (yr)
Srage of handicap ar the
time of examination
Clinical symptoms
Cerebellar syndrome
Reflexes lower limbs
Babinski sign
Decreased vibration sense
Dysronia
Ophrhalmoplegia
Fasciculations
Intellectual impairmenr
Axonal neuropathy

SCAl

SAL-318-008

SAL-303-003

SAL-303-006

SAL-326-038

Woman
82
22
26
16
38
Wheelchair bound

Woman
74
44
56
13
57
Wheelchair bound

Woman

60
67

Man
56
32
37

11
71
Walked with help

42
Wheelchair bound

+
+

Increased

Increased

++
+

++
+

10

t+

Abolished

Increased

N 11
-

ND
Blepharospasm

Tongue

ND

++
+

+
+

++

+
+

SCA = spinocerebellar ataxia; MJD = Machado-Joseph disease; NI> = not derermined.

patients and in the 2 SCAl patients but slightly longer

in the former than in the latter. No clinical syniptom
clearly distinguished SCA31MJD patients from those
with SCAl, although ophthalmoplegia was more pronounced in the former and intellectual impairment in
the latter. Individual SAL-318-008 (SCA31MJD) and
SAL-303-006 (SCA1)were idiosyncratic because of the
presence of severe dystonia in the former, and abolished lower limb reflexes in the latter.
Some neuropathological features of the two diseases
were distinctive (Fig 4). In general, basal ganglia lesions
were more severe in SCA3/MJD than in SCAl (Table
5). In particular, the internal pallidum and the subthalamic nucleus, which were more severely degenerated
in SCAS/MJD, were scarcely or not at all affected in
SCAl. On the contrary, olivopontine and cerebellocor-

tical lesions were more severe in SCA1, where Purkinje

cell loss and inferior peduncle degeneration were found
exclusively. Degeneration of efferent cerebellar systems
(dentate nucleus, superior cerebellar peduncle) was seen
in both SCAl and SCA3. Spinal lesions were found
in both diseases but affected different regions; the posterior columns were more affected in SCA1, the
Clarkes columns in SCA3/MJD, and the intermediolateral column only in the latter.
Marked individual variations were observed, eg, severe oculomotor and corticospinal tract degeneration
in SCA3/MJD Patient SAL-309-003. Patient SAL309-003, with 74 CAG repeats, had severe degeneration of the posterior column, whereas Patient SAL318-008, with 82 CAG repeats, did not. This reflects
clinical observations that the frequency of decreased vi-

Diirr et al: SCA3 and MJD

495

Fig 4. Neriropirtliology. (A) lnf;.rior oliue o f Case SAL-303-006 (spinocerebel/ar ataxin I [SCAI]). Bodian stntn i.otcpkd iuitb
liixol jist blzre. ( X 20 befwe 36% reduction.) (B) htfcrior olive of case SAL-318-008 (SCA3). Bodinn stain coupled with luxol
jiit blur. ( X 20 b4ore 36Yo mfiictiori.) 111 A, there ioizs a marked loss qf'neurons responsible f o r the ntrophy and pallor of the
hilus (arrow), to be conipnred iclith the i i o r m d ~stained
~~
hilus (arrow) in B. (C) Enipt?, briskets (arrows) due to loss of Purkinje
cells in c m SAL-326-038 (.SCAI). Bo&v strEiii coupled with Ii~uolfdSt blue. ( X 400 before 35% reduction.) (0)Norniril
Purki)!jc i d i (arrows) in cizse SAL-.309-003 (SCA3). Bodiirri stain coupled with luxol $st blue. ( X 400 before 36% reduction. )
bration sense is negatively correlated with CAG repeat
numbers (see Table 3).
Clinicopathological correlations in SCAS/MJD
raised several issues; ie, both patients had severe basal
ganglia atrophy, but only SAL-318-008 had dystonia,
and both had lesions of the intermediolateral column,
but neither had clinical symptoms of dysautonomia.

Discussion
Previously, we showed that both SCA3 and MJD patients carry expanded alleles in the MJDI gene (161.
This finding justifies the inclusion of MJD in the clinically and genetically heterogeneous group of ADCA
type I, as proposed by Harding [l]. The CAG expansion at the SCA3/MJD locus constitutes a major cause
of ADCA type I, representing 28941 of the families;
whereas SCA1 accounts for only 13"/0 in our subset of
families. A similar frequency for SCA3/MJD is re-

496

Annals of Neurology

Vol 39

No

4 April 1996

ported for a smaller series of German ADCA I families

[15]. Like SCA1, the SCA3/MJD mutation is found
exclusively in ADCA type I patients, never in those
with ADCA type 11 or 111, or in either isolated o r recessively inherited cases of olivopontocerebelIar atrophy

[6, 71.
T h e analysis of 126 gene carriers from 34 families
permits characterization of the behavior of the CAC,
expansion. T h e numbers of CAG repeats are clearly
different on normal (14-40) and expanded alleles (6482). A 21 CAG repeat difference remains when the
largest normal allele (40) and the smallest expanded
allele (61) reported in the literature I201 are compared.
The distribution of the normal alleles varies among
populations; ie, the major allele carries 23 repeats in
the French, Brazilian [21], German [ I 51, and Anierican
populations [22] but 14 in the Japanese [14, 20, 231.
There is a strong negative correlation between the

Table 5. Neuvopatbological Findings

SCASI M JD

Cerebral cortex
External pallidum
Internal pallidum
Subthalamic nucleus
Substantia nigra
Red nucleus
Oculomotor neive
Locus coeruleus
Pontine nuclei
Medial longitudinal fasciculus
Inferior olives
Twelfth nerve
Superior cerebellar peduncle
Medial cerebellar peduncle
Inferior cerebellar peduncle
Purkinje cells
Dentate nucleus
Spinocerebellar tracts
Corticospinal tracts
Posterior columns
Intermediolateral column
Clarkes column
Anterior horns
SCA

= spinocerebellar ataxia;

MJD

SCA 1

SAL-3 18-008

SAL-309-003

SAL-303-006

SAL-326-038

+
++
++
+
++
+
+

+
++
++
+
+
+++
+
++
+

+
-

++
+

++
+++

+
-

++

+
++
+++

+++
+

++
+

++
++
++
+
+
++

++
+
++
++
+++
++

++

++
++
-

+
+++
+

++
+++
+
++
++
+++
++

++/+++
-

+
+

-t

++
++
+

= Machado-Joseph disease.

number of CAG repeats on the mutated allele and age

at onset ( 7 = -0.8, p < O.OOI), similar to other reported values [13-15, 20, 22, 231. We provide, however, the first evidence of a small but significant contribution of the normal allele ( r = -0.25, p < 0.05).
The combined effects of the normal and the expanded
alleles account for 70% of age at onset variability. The
remaining 30% is determined by unknown factors.
The variation, which was larger among than within
families, indicates that other genetic factors might play
a role.
The mutation is unstable when transmitted through
both male and female germ lines. Expansions and contractions were observed in both sexes. The observation
of larger as well as smaller expanded alleles in sperm
DNA compared with blood [16] may explain the occurrence of both expansions and contractions during
male transmission. In this study, although the mean
increase in the number of CAG repeats observed in
parent-child couples was larger in paternal (+ 1.4)
than in maternal transmission (+0.5), the difference
was not statistically significant. Paternal transmissions
seem to be more unstable and tend to be expanded
[13, 231 but much less markedly than in Huntingtons

disease (HD) [24], DRPLA [25], and SCAl [7].

In
DRPLA, the repeat length always increases during male
transmission and large expansions of more than 20 repeats have been documented in HD, DRPLA, and
SCAl but not in SCA3/MJD [24-261. In addition,
we showed that instability is not influenced by either
the size of the CAG repeat on the transmitted expanded allele or by the age at conception.
Anticipation exists in families with the SCA3/MJD
mutation, as a consequence of the observed mean increase of nearly one CAG repeat from one generation
to the other. However, the apparent anticipation in
parent-child pairs is larger than expected from the regression between age at onset and number of CAG
repeats. This is due to an observation bias, however,
since only living parent-child pairs were included in
which the onset is likely to be earlier in the child than
in the still-living parent.
The wide interindividual clinical variability is determined by the number of CAG repeats and disease duration. The frequencies of abnormal reflexes, vibration
sense, and probably the presence of axonal neuropathy
are correlated with the number of CAG repeats, resulting in what were thought to be three clinical sub-

Durr et al: SCAS and M J D 497

types of MJD, apparently determined by age a t onset

[2, 20, 271.However, none of these correlations is perfect, and the interindividual variation cannot be explained solely by the size of the expanded CAG repeat. Supranuclear ophthalnioplegia, a.myotrophy, and
sphincter or swallowing difficulties. lor example, are
not related to the number of CAG repeats but increase
with disease duration, as previously shown in a large
series of patients with ADCA type I 1281.
We have previously shown differences in the frequencies of the signs associated with cerebellar ataxia
in SCA3 compared with those in a large series of Azorean MJD patients [ 161. Since the sizes of the expanded
alleles and disease durations are not known in the latter, no definite conclusions can he drawn. Bulging eyes
and early extrapyramidal signs including dystonia are
considered t o be characteristic of MJD but not SCA3
[ 2 2 ] . However, these signs were as frequent in SCAl
as i n SCA3 patients in this study.
The overall clinical pictures of patients with mutations at the SCAl and SCA3/iMJD locus are very siniilar. This might explain why Japanese patients with
clinical MJD did nor carry the SCA3/MJD expansion
[14. 20, 231. These patients may carry the SCAl mutation, or another genetic anomaly responsible for
ADCA type I.
Significant group differences are seen between
SCA3/MJD and SCA2 patients but do not concern
the signs that are related to disease dwation. The presence of signs such as decreased reflexes and dementia,
which are globally niore frequent in SCA2, is not sufficient, however, to guarantee the diaignosis of SCA2
in individual patients. For these reasons, we believe
that only molecular analysis can determine the final
diagnosis in patients with ADCA type I.
Major neuropathological features clearly distinguish
the SCAl from SCA3/MJD cases, internal pallidum
and intermediolateral colutnn lesions in SCA3/MJD,
inferior olive and Purkinje cell degeneration in SCA1.
It has been stated that neurodegeneration in SCA3/
MJD is more homogeneous than in SCAl or 2 and
that degeneration of the subthalamopallidal system is
not present in the latter (91. In contradiction to this
notion, on the one hand, the subthalamic nucleus was,
although slightly, affected in one of our SCAl patients
with neuropathological examinations; on the other
hand, the oculoniotor nerve was degenerated in I of
the SCA3/MJD patients. The severity of some neuropathological lesions in SCA3/MJD patients could be
related to the length of the CAC; repeat on their niutated allele. or to somatic mosaicisni.
In conclusion, SCA3/MJD represents, so far, [he
most frequent unstable mutation responsible for
ADCA type I. We show that the number of CAG repeats determines not only the age at onset, but also to
some extent the clinical presentation. Disease duration,

498

Annals of Nrurologv

Vol 30

Nv 4

April 1996

however, also contributes to the interindividual clinical

variability. T h e instability during paternal transmission
of the SCA3/MJD locus is less marked than in other
neurodegenerative disorders caused by expanded CAG
repeats, suggesting the role of unknown cis-acting factors.

We thank the VERUM foundation, the Association Frnnfaise contre les Myopathies, I he Groupement de Recherches et dEtudes bur
les Ginomcs, and the Association pour l e Developpenienr de la
Rzcherche sur les Maladies Ginttiques Neurologiques et Psychiatriques for financial support.
W e are grareful to the families for participating and we thank Drs
S. Backchine, M. Bcquet, J.-J. Cassinian, P. Damier, 1. Emile, G .
Finelon, 3. Fontaine, D. Grid, M. Haguenau. hl. Koenig, A. D.
Korayn, P. Labaugc, D. Laplane, C. Legum, J:F. Prudhornme,
A. Toutain. and T. Uchihara. W e thank Y. Pothin, J. Rou, and C .
Ponthieu for technical assistance. Many thanks t o Merle Ruberg for
critical reading of the manuscript.

References
I . Harding AE. Aurosomal dominant cerebellar ataxia of late onset. The heredirary ataxias and related disorders. New York:
Churchill Livingsrone 1984:123- I65
2. Rosenherg RN. Autosomal dominant cerebellar phenotypes:
the genotype has settled the issue. Neurology 1995;45:1-5
3. Yakura H, Wakisaka A. Fujimoto S, et al. Hereditary ataxia
and H I A genotypes. N Engl J Med 1974;291:154-155
4 . Orr HT, Chung M-Y. Banfi S, et al. Expansion of an unstatile
trinucleoride CAG reprat in spinocerebellar araxia type 1 . Na1-226
ture Genet 199.$;4:22
5 . Ranum I.PW, Chung M, Banfi S. et al. Molecular and clinical
correlations in spinocerebellar ataxia 1 (SCA I ) : evidence for
familial effects on the age at onset. Am J Hum Genet 1994:

55:244-252
6. Giunti P, Swceny M G , Spadarn M, et 31. The trinucleotide
repeat expansion on chromosome 6p (SCA1) i n autosonial
dominant cerebellar ataxias. Brain 1994;1 17:645-649
7 . Dubourg 0 , Diirr A, Cancel G. et 31. Analysis of the SCAl
CAG repeat in a large numher of families with dominant
araxia: clinical and molecular correlations. Ann Neurol 1995;
37:176- 180
X. Gispert S, Twells R, Orozco G. er al. Chromosonial assignment of the second locus for autosonial dominant cerebellar
ataxia (SCA2) to human chromosome 12q23-24.1. Nature
Genet 1993;4:2:)5-299
9. Takiyanu Y, Nishizawa M , Tanaka H, et al. The gene for
Machado-Joseph disease maps to chromosome 14q. Nature
Genet 1993;4:3OO-304
10. Stevanin G. Le Guern E. Kavisi N,et al. A third locus for
nurosomal dominant cerebellar ataxia type I maps to chromosome 14q24.3-qrer: evidence for the existence of a fourth locus.
Am J H u m Genet 1994;54:11-20
1 1 . Srrvanin G , Sousa PS, Cancel G. ct al. The gene for MachadoJoseph disease maps to the same 3-cM interval as the spinal
cerebellar ataxia 3 gene on chromosome 14q. Ncurobiol Dis
1934;1:79-82
12. Twist EC, Casaubon LK, Ruttlcdge H , et al. Machado-Joseph
disease maps to the same region of chromosome 14 as the
spinoccrebellar ataxia type 3 locus. J Med Genet 1394;32:25.3 1
13. Srevanin G. Cancel G, D ~ r A.
r et d.The gene for spinal cerebcllar ataxia 3 (SCA3) is located in .I region of -3chl on

14.

15.

16.

17.

18.

19.

20.

21.

chromosome 14q24.3-q.32.2. Am J H u m Genet 199556: 193201

Kawaguchi Y. Okarnoto T, Taniwaki M, et al. CAG expansions in a novel gene for Machado-Joseph disease at chromosome 14q32.1. Nature Genet 1994:8:221-228
Schols L, Vieira-Saecker AMM, Schiils S. et al. Trinucleotide
expansion within the M J D l gene presents clinically as spinocerebellar ataxia and occurs most frequently in German SCA
patients. Hum Mol Genet 1395;4: 1001-1 005
Cancel G. Ahbas N , Stevanin G , et al. Marked phenotypic
heterogeneity associated with expansion of a CAG repeat sequence at the spinocerebellar ataxia 3/Machado-Joseph disease
locus. Am J H u m Genet 1995;57:809-816
Gardner K , Alderson K, Galster B, et al. Autosomal dominant
spinocerebellar ataxia: clinical description of a distinct hereditary ataxia and genetic localization to chromosome 16 (SCA4)
in a Utah kindred. Neurology 1394;44(suppl 2):921S
Ranuni LPW, Schut LJ, Lundgren J. et al. Spinocerehellar
ataxia type 5 in a family descended from the grandparents of
President Lincoln maps to chromosome 11. Nature Genet
1974;8:280-284
Diirr A, Smadja D, Cancel G, et al. Autosomal dominant cerebellar ataxia type 1 in Martinique (French West Indies): clinical
and neuropathological analysis of 53 patients from three unrelated SCA2 families. Brain 1935;118:1573-1581
Maruyama H , Nakaniura S, Matsuyama Z, et al. Molecular
features of the CAG repeats and clinical manifestation of Machado-Joseph disease. H u m Mol Genet 1395;4:807-8 12
Srevanin G, Cassa E, Cancel G, et al. Characterisation of the

22.

23.

24.

25.

26.

27.

unstable expanded CAG repeat in the M j D l gene in four Brazilian families of Iortuguese descent with Machado-Joseph disease. J Med Genet 1995;32:827-830
Matilla T, McCall A, Subramony SH, Zoghbi HY. Molecular
and clinical correlations in spinocerebellar ataxia type 3 and
Machado-Joseph disease. Ann Neurol 1995;38:68-72
Takiyama Y, Igarashi S, Rogaeva EA, et al. Evidence for intergenerational instability in the CAG repeat i n the M J D l gene
and for conserved haplotypes at flanking markers amongst Japanese and Caucasian subjects with Machado-Joseph disease.
H u m Mol Genet 1995;4:1 137-1146
Telenius H , Almqvist E, Krenier B, et al. Somatic mosaicism
in sperm is associated with intergenerational (CAG)n changes
in Huntington disease. H u m Mol Genet 1995;4:189-195
Ikeuchi T, Koide R, Tanaka H. et al. Dentatorubral-pallidoluysian atrophy: clinical features are closely related to unstable
expansions of trinucleotide (CAG) repeat. Ann Neurol 1995;
37:769-775
Chung M. Ranum L, Duvick L. et al. Evidence for mechanism
predisposing to intergenerational CAG repeat instability in spinocerehellar ataxia type I. Nature Genet 1993;5:254-258
Coutinho P, Andrade C. Autosomal dominant system degeneration i n Portugucse families of the Azores Islands: a new genetic disorder involving cerebellar, pyramidal. extrapyramidal
and spinal cord motor functions. Neurology 1978;28:703-

709
28. Durr A, Chneiwciss H, Khati C, et al. Phenotypic variability
in autosomal dominant cerebellar ataxia type I is unrelated to
genetic heterogencity. Brain 1993; 1 16: 1497- 1508

Diirr et

a]:

SCA3 and MJD

499