Don
A.
Wilson
1
John R. Prince
This article appears in the March/April 1989
issue
of
AJNR
and the May 1989 issue
of
AJR.
Recipient of the John Caffey Award for Best
Scientific
Paper presented at the annual meeting of
the Society for
Pediatric Radiology,
San
Diego,
1988.
'Both
authors: Magnetic Resonance Center,
Oklahoma Medical Center, 940
NE
13th St., P.O.
Bo
x 26307, Oklahoma City, OK 73126. Address
reprint requests to
D.
A. Wilson.
AJNR
10
:259-262,
March/
April 1989
0195-6108/
89
/1002-0259
© American Society of Neuroradiology
MR Imaging Determination of the
Location of the
Normal Conus
Medullaris Throughout Childhood
259
This retrospective study was designed
to
determine the location
of
the conus med-
ullaris in normal children
by
reviewing a series
of
MR
images
of
the lumbar spine. The
study
group
consisted
of
184 children ranging in
age
from newborn
to
20 years who had
a
normal conus level as reported by the radiologist of record. The range of conus levels
for
the entire
group
of
normal children was T12
to
L3. The range
for
the
0-2-year-old
group was T12 to
L2-L3
with an average
of
L
1-L2.
The range
of
conus levels for the
19-20-year-old
group
was
L 1 to L2 with an average
of
L
1-L2.
We
conclude
that the conus medullaris
does
not ascend throughout childhood as
stated
by
previous authors
but
attains the adult level sometime during the first few
months
of
life. A conus level
at
L2-L3
or
above should
be
considered normal
at
any
age. A conus
level
at
L3 is indeterminate, since
it
is
possible
for
a normal
or
a tethered
conus
to
be located
at
this level.
In
order to diagnose a low-lying or tethered cord
in
children, it
is
necessary to
know the
level at which the normal conus medullaris terminates. This
is
a topic
that has not received a great
deal of attention in the literature. Commonly used
reference texts for pediatric neuroradiology and pediatric imaging
[1
, 2] state that
the
lower level of the conus
is
normally at the third lumbar vertebral body (L3) at
birth and that it reaches
L2-L3
by age
5,
L2
by age 12, and eventually arr
iv
es
at
about L 1
by
adulthood. This series of age-related levels clearly suggests that the
conus ascends throughout
childhood and that different criter
ia
of normality should
be
used for different ages. The purpose of this study was to develop· a better
statistical basis for determining the location of t
he
normal conus medullaris through-
out
childhood by reviewing a large seri
es
of
MR
images. Comparisons were also
made with a group of children who had surgically proved tethered cord syndrome.
Materials and
Methods
MR imag
es
of the lumbar spine were retrospectively reviewed
in
184 children ranging
in
age from newborn to 20 years. The subjects were divided into
two
groups. Group I consisted
of 85 infants
and
children (42 girls
and
43 boys) who were studied for reasons other than
sp
in
al
dysraphism. These patients were referred for
MR
imaging because of low back pain
or back injury (37, or 43%
),
possi
ble
infecti
on
or tumor of the spine or cord
(21
, or 25%),
brain tumor to
exclude dropped metastasis
(11
, or 13%), leukemia
(11
,
or
13%),
and
abdominal mass
(5,
or 6%
).
The average age of this group was 1 0.8 years.
Group
II
contained 99 subjects (50 femal
es
and 49 males) referred with possible symptoms
of tethered cord syndrome but reported to be
normal by the radiologist of record. The clinical
in
dications for these studies were
as
follows: poor bow
el
and/or bladder control
(3
9,
or 39
%),
scoliosis (27 , or
27
%), abnormal muscle tone or gait (25, or 25
%),
spina bifidia (13, or 13%),
foot deformities (13, or 13%), and
sacral dimple or hairy patch
(8
, or 8%). The clinical
indications exceed the total number of patients
in
the group because many of the patients
had
multiple symptoms. The
ag
e range for this group was 5 days to 20 years, with
an
av
erage
age of 7.9 years.
Over the same time span
in
which the normal subjects we
re
be
in
g studied (approximate
ly
260 WILSON AND PRINCE
AJNR
:10, March/April1989
3 years), 40 infants and children (25 girls and 15 boys) from our
in
stitution were diagnosed and surgically confirmed to have the
te
th
ered cord syndrome. These children ranged
in
age from birth to
19 years, with
an
average age
of
6.7 years. These cases were
reviewed to determine the range of
pathologic conus levels
in
our
population.
MR
s
can
s
of
the lumbar spine were also reviewed
in
1 00 young
adults who were referred for lumbar disk disease but who did not
have symptoms or signs of
myelodysplasia. This group was used to
establish the MR criteria for the normal adult conus level. There were
37 women and 63 men ranging
in
age from
21
to 40 years. The
average age of this group was 32 years.
All
the imaging studies for both groups were obtained on super-
conducting magnets operating at
0.5 T* or 1.5 T.t To be included
in
this retrospective investigation,
an
MR
study had to consist of a
good-quality sagittal T1-weighted image of the lumbar spine that
included visualization of the distal spinal cord and conus medullaris.
The pulse sequences used were
350-600/20-30/2-4
(TR
range/TE
range/number of excitations). The
slice thickness was 3 mm for
62/
85 (73%)
of
the subjects
in
group I and 68/99 (69%)
of
the subjects
in
group
II.
The remainder of the examinations were done at 5-mm
slice thickness. The matrix was 128 x 256
or
256 x 256, with a
20-
40-cm field of view. The level
of
the conus medullaris was determined
in
each instance by first locating the lumbosacral junction on the
sagittal MR image and assuming that there were five lumbar vertebral
bodies. The intervertebral disk space
or
midpoint of the vertebral
body closest to the tip of the conus was recorded as the conus level.
Coronal
T1-weighted images were also obtained by using the same
pulse sequences cited above for the sagittal images
in
52/85
(61
%)
of
group-1
subjects and
in
92/99 (93%)
of
group-11
subjects. The
coronal images were correlated with the sagittal images to confirm
the
le
vel
of conus termination (Fig . 1
).
Some
of
the subjects also had
balanced and T2-weighted sagittal images with parameters
of
2000-
21
00/20-30,80-100/2
. These images were used for additional con-
firmation
of
the findings on the T1-weighted images.
In
the total study
• S5 Gyroscan (Philips, Shelton,
CT)
.
t S15 Gyroscan (Philips, Shelton, CT) or Signa G. E
.,
Milwaukee,
WI).
A
8
group of 184 subjects, only
11
(6
%) who were imaged at a 5-mm
slice thickness did not have coronal images.
All
of the lumbar spine
MR
images were obtained by using rectangular surface coils with
dimensions of
10
em
x 20
em
or 10
em
x 40
em
(Philips) or 13
em
X 28
em
(GE).
Radiographs of the lumbar spine were available for review
in
46/
85 (54%) of the
group-1
subjects and
in
50/99
(51
%) of the group-
II
subjects. The true number of lumbar vertebrae and the occurrence
of
transitional vertebrae were noted
in
each case and adjustments
were made to the designated
level of conus termination
as
necessary.
For
statistical analysis, the anatomic level determined for each
conus was converted to
ordinal data using the following schema. The
84
through
81
vertebral bodies were assigned a score of 1 through
4,
respectively. The L5/S1 disk space was assigned a score of 5,
L5
a score
of
6,
the
L4-L5
disk space a score
of
7, and so on through
T11
, which was assigned a score of 18. The means and standard
deviations were determined by age
intervals for both groups. These
ranking scores were further
analyzed for statistical differences be-
tween groups
as
a whole and by age intervals using the Mann-
Whitney test.
Results
The range of conus levels,
in
brackets, and the mean score
± 1 standard deviation,
in
parentheses, by age intervals
rounded to the nearest year tor
group-1
subjects (normal
controls)
were:
0-2
years (T12.5*-L 1.5](13.
8§
± 0.57),
3-4
years (T12.5-L3](13.6 ± 1.58),
5-6
years (L 1- L2](13.4 ±
0.89),
7-8
years
[T12.5-L
1 ](14.1 ± 0.38),
9-10
years [L
1-
L2](13.3 ± 1.15),
11-12
years [L 1.5](13),
13-14
years [L
1-
L 1.5](13.2 ± 0.44),
15-16
years (L
1-L
1.5](13.6 ± 0.53),
17-
18 years [T12.
5-L
1.5](13.8 ± 0.75),
19-20
years [L
1-L2]
(13.1
± 0.78).
; T12.5 indicates the
T12-L
1 disk space, etc.
§ 13 indicates the L
1-L2
disk space, 14 indicates the L 1 vertebral body.
Fig.
1.-Sagittal
(A) and coronal
(8)
MR
im-
ages
(600/20)
show
good
correlation
of
conus
location
at
L 1 in this normal 2-year-old girl.
The range of conus levels,
in
brackets, and the mean ± 1
standard deviation,
in
parentheses, by age intervals rounded
off to the nearest year for
group-11
subjects (patients with
symptoms but whose
MR
findings were normal) were:
0-2
years [T12-L2.5](13 ± 1.45),
3-4
years [T12.5-L2.5](13.5 ±
1
.3
0)
,
5-6
years [T12.5-L2](13.2 ± 0.97),
7-8
years
[T12-
L2](14.1 ± 1.46),
9-10
years [T12-L2.5](13.2 ± 1 .48),
11-
12
years [L1-L2](12.9 ± 0.78),
13-14
years
[T12-L2
.5](12.8
± 1.48),
15-16
years [L1-L2.5](12.0 ± 1.73),
17-18
years
[T12.5-L2](13.0
± 1.22),
19-20
years [L
1-L
1.5](13.7 ±
0.
58).
The mean conus level for groups I and
II
was compared
and
found to
be
the same at each age interval at a confidence
level
of 95
%.
Consequently, groups I and
II
were compared
as
a whole. The means and standard deviations were 13
.5
±
0.85 and 13.2 ± 1.32 for groups I and
II
, respectively. There
was no statistical difference between these groups at a
confidence level of 95
%.
These two groups were therefore
combined, giving a grand mean and standard deviation for
both groups of 13.3
± 1.13. This corresponds to
an
approx-
imate anatomic level of the L
1-L2
intervertebral disk space
and
two standard deviations would
be
T12 to
L2-L3
. The
ranges and averages for the combined groups are given
in
Figure 2.
The conus level
in
the 1 00 normal adults ranged from
T11-
T12 to
L2-L3
with the average level at L
1.
The patients with surgically proved tethered cords had
conus levels ranging from
L3
to
S4
, with the average level of
termination at
L4-L5.
No age-related pattern was observed
in
these patients.
There were radiographs of the lumbosacral spine available
for review on 96/184 (52%) of the group
I and group
II
subjects.
We
found eight children with transitional vertebrae,
two children with four lumbar vertebral bodies, and
two
children with six lumbar vertebral bodies. The appropriate
corrections were made
in
each instance to reflect the true
level
of conus termination. The conus level of termination was
corroborated by myelography
in
six cases from group I and
in
five cases from group
II.
Discussion
The volume of data used to develop the present-day criteria
for the normal conus level
in
children
is
in
fact rather small.
V>
·
.::
""
T11
::l
'0
There are three anatomic studies of children
in
the recent
literature that evaluated conus level by the dissection of
cadavers. The most widely quoted study,
by
Barson [3]
in
1969,
is
actually a study of cord levels
in
the fetus (only 12
specimens over the age of 1 year were examined).
He
dem-
onstrated conclusively that the conus level does ascend
throughout fetal life, but, on the basis of data from a very
small number of infants and children, suggested that the adult
level of L
1-L2
was attained
by
the age of 8 weeks.
In
another
anatomic study, Jit and Charnalia [ 4] reported the average
location of the tip of the conus
in
1 0 newborn infants
as
the
middle of
L2
, with a range from L
1-L2
to
L2-L3
. No older
children were studied. These researchers accepted the adult
level
as
L
1-L2
and noted that there was a change of only half
a vertebral body from birth to adulthood. The other anatomic
study with information pertinent to this problem was published
in
1972
by
James and Lassman [5], who examined the conus
level
in
autopsies of 25 children ranging
in
age from newborn
to 8 years. They reported that the conus reached the
L2
level
no
later than the age of 5 months, but they examined only
three specimens over the age of
1 year.
The combined anatomic data from these three studies
contain or.ly 15 measurements
in
children over the age of 1
year. This
is
insufficient evidence to justify the accepted level
of termination of the normal conus throughout childhood.
No
imaging studies of significant size have been performed to
better define the level of the normal conus
in
children, even
though it
is
well accepted that myelography
[1
, 6] and
MR
imaging [7]
can
accurately make this determination.
Our study contradicts the conventional wisdom that the
conus
medullaris ascends throughout childhood.
We
found
that there is
no
significant difference
in
the level of conus
termination among the various age groups of normal children
(see
Fig
.
2).
At any age from birth to 20 years, a conus level
of termination at or above
L2-L3
was normal. The subjects
with a conus level at the L3 level provided some difficulty.
We
encountered one normal 3-year-old girl with a conus at this
level
(Fig
.
3)
. There were also
two
patients with surgically
proved tethered cords that terminated at this level. This
is
not
surprising when we note that Reimann and Anson [8] deter-
mined from their study and review of
801
adult cadavers that
1.8% of normal adults
will have a conus level of termination
at L3.
--
~
----
-
t
---
-l
---
-t
-----
-----
r
---
-----------±-------
--
-
f ! -
-
"'
T1
2
:2
V>
L1
1!
0
u
L2
Fig.
2.-Average
conus
level
(dots)
and range
(vertical bars) are indicated
for
184 normal chil-
dren, ages newborn
to
20 years, and for 100
adults.
0
.
9-
0
a;
>
"'
-'
1
L3
L4
L5
[31]
0
-
[25] [23]
[14]
[12]
[11] [21]
[10]
[17]
[20]
(1
00]
2 3 4 5 6 7 8 9 10
11
12 13 14 15 16 17 18 19
20
Adul1
Ag
e in Years
262
WILSON
AND
PRINCE
AJNR
:10, March/April1989
Fig.
3.-Sagittal
MR
image
(450/30)
of
lumbar
spine in normal
3-year
-
old girl with a conus tip located at
mid-body
of
L3. 3 = L3 vertebral body.
The possibility that the conus ascends from birth to the
adult level by age 8 weeks
as
reported by Barson [3] could
not
be
confirmed from our data. Only 20 of our subjects were
under 1 year
old, and of these only six were age 8 weeks or
younger.
In
this small group of subjects, the range of conus
levels was
T12-L
1 to
L2-L3
, with
an
average level at L
1-L2
,
results very similar to Jit and Charnalia's [4] group of 10
newborns. Additional study of this early age group by
MR
imaging seems warranted.
MR
imaging appears to
be
an
accurate method of deter-
mining the level of conus termination. Myelography agreed
with
MR
in
the
11
cases
in
which both studies were per-
formed. Additional myelographic confirmation was difficult to
obtain because of the high
level of confidence
in
our results
reported by referring physicians.
In
the large adult study
reported by Reimann and Anson [8], the average
level of
termination of the
adult conus was the lower one-third of L 1.
The same
conclusion was also derived from our series of 1 00
adult
subjects.
The
coronal T1-weighted images available
in
144/184
(78%) of our subjects did not
significantly change the level of
conus termination
as
determined
on
the sagittal images. The
coronal images were definitely useful
in
increas
in
g our confi-
dence that we were indeed correctly identifying the conus
in
patients with moderate to severe scoliosis. Correlation of
sagittal images with coronal images
is
essential
in
patients
with
an
abnormal conus or thickened filum terminale but
would appear to
be
less important
in
normal individuals.
The degree of potential error introduced into this study by
the
possible occurrence of undetected transitional vertebrae
or the presence of four or six
lumbar vertebrae
is
considered
negligible. Ford
and
Goodman [9]
in
a review of 1614 lum-
bosacral
spine films determined the incidence of transitional
vertebrae to
be
about 8%, with four lumbar vertebral bodies
occurring
in
4.2% of subjects and six lumbar vertebral bodies
occurring
in
3% of subjects.
We
corrected for this occurrence
in
over half
(52
%) of our patients. For the 88 patients without
lumbosacral spine radiographs, the expected incidence of
transitional vertebrae would
be
seven subjects, the incidence
of four
lumbar vertebral bodies wou
ld
be
four subjects, and
the incidence of six
lumbar vertebral bodies would
be
three
subjects.
We
conclude that the conus medullaris does not ascend
throughout
childhood
as
suggested by many authors, but
attains the
adult level sometime during the first few months
of
life. Further study of this young age group
is
warranted. A
conus
level of termination at
L2-L3
or above
is
normal at any
age. A conus
level of termination at
L3-L4
or below
is
abnormal (with possible exceptions occurring
in
premature
infants and
full-term newborns). The normality
of
a conus
medullaris located at
L3
must be determined by other means.
ACKNOWLEDGMENTS
The authors are grateful to
Pat Barnes and
Dan
Galloway (de-
ceased) for the early work they did at Oklahoma Diagnostic Imaging,
Inc.
,
in
setting up the original spine imaging techniques. Their efforts
enhanced our
abi
lity to produce the superb quality images that were
reviewed for this study.
We
also thank the radiologists of record
on
the cases
in
our study, Pat Barnes, Nguyen
Dan
, Susan Edwards,
Dan
Galloway, Edmond Kalmon, Sidney Traub, and Max Walter, for
their significant contributions.
We
appreciate the support
and
cooperation we received from the
Board of Directors and the staff of
Oklahoma Diagnostic Imaging,
Inc.
, for allowing us to use some of their case material
in
our project,
and
we are grateful for the assistance of Glenda Sims, Mary Barnard,
and
Julie Wilson
in
the preparation of this manuscript.
REFERENCES
1. Harwood-Nash
DC
, Fitz
CR
. Myelography.
In
: Neuroradiology
in
infants
and children.
St. Louis: Mos
by
, 1976:1133
2. Kirks
DR
. Spine
and
contents.
In:
Practical pediatric imaging. Boston/
Toronto: Littl
e,
Brown, 1984:134
3. Bars
on
AJ
. The vertebral level of termination of the spinal cord during
normal a
nd
abnormal development. J Anat 1969;1 06:
489-497
4.
Jit I,
Ch
arnalia
VM
. The vertebral
le
v
el
of the termination of the spinal cord.
J Anat Soc India 1959;8:
93-102
5. James CCM, Lass
man
LP
. Spinal dysraphism.
In
: Spina bifida occulta.
London: Butterworth, 1972:20-21
6. Fitz
CR
, Harwood-Nash
DC
. The tethered cord. AJR 1975;
125:515-523
7.
Barn
es
PO
, L
es
ter
PO
, Yamanashi
WS
, Prince
JR
. Magnetic resonance
im
a
gin
g
in
infants
and
children with spinal dysraphism. AJNR
1986;7:465-
47
2,
AJR 1986;147:
339-
346
8.
Reim
a
nn
AF
, Anson
BJ.
Vertebral le
vel
of termination of the spinal cord
with report of a
ca
se
of sac
ral
cord. Anat
Re
c 1944;88:
127-138
9. Ford L T, Goodman
FG
.
X-
ray studies of the lumbosacral spine. South Med
J 1966;59:
11
23-11 28
