The charts and radiographs of twenty-five patients with partial or
total absence of the lumbar spine and total absence of the sacrum
who had been evaluated at our institution between 1951 and 1998
were reviewed. All of the patients had been examined by the senior
author (S.J.K.) in a spinal dysfunction clinic. Seven patients were
seen only once in consultation. The other eighteen patients, who
had adequate follow-up, were included in this study. They were divided
into two groups (Table I). Group I consisted of thirteen
patients with only lumbosacral agenesis, and Group II consisted
of five patients with lumbosacral agenesis and a myelomeningocele.
We recorded whether there was a history of maternal diabetes mellitus;
the sensory and motor levels; and the nature of any hip dysplasia, lower-limb
deformities and contractures, scoliosis, or cervical spine anomalies.
All previously performed orthopaedic procedures were also noted.
Visceral and other nonmusculoskeletal congenital conditions were
noted as well but were not used in the data analyses as they were
not pertinent to the study.
The criteria of Hoffer et al.9 were
utilized to categorize the ambulatory status of these individuals.
Functional ambulators were divided into two subgroups: community
ambulators and household ambulators. Community ambulators could
walk indoors and outdoors for most of their activities. They used
crutches or braces, or both, and they needed a wheelchair only for
long trips. Household ambulators could walk only indoors. They required
assistive devices but could transfer themselves in and out of bed
with little or no assistance. They used a wheelchair for some indoor activities
and for all activities in the community. Nonfunctional ambulators
could walk only during physical therapy sessions, and nonambulators
required a wheelchair at all times but could transfer from chair
to bed.
Initially, we tried to use the classification of lumbosacral agenesis
described by Renshaw4, but we
could not adapt it to our patient population as the population included
only patients with total or partial lumbar agenesis and complete
sacral agenesis. We defined complete sacral agenesis as the absence
of bone or the presence of only a small segment with no structural
pattern caudad to the lumbar spine. We therefore developed a different
classification system that encompassed this group. The spinal deformities were
divided into three types. In Type A, there was either a slight gap
between the ilia or the ilia were fused in the midline. One or more
lumbar vertebrae were absent. The caudad aspect of the spine articulated
with the pelvis in the midline, maintaining the vertical alignment
of the spine (Fig. 1). In Type B, the ilia were fused
together, some of the lumbar vertebrae were absent, and the most
caudad lumbar vertebra articulated with one of the ilia, with the
most caudad aspect of the spine shifted away from the midline (Fig. 2). In Type C,
there was a total agenesis of the lumbar spine, the ilia were fused
together, and there was a visible gap between the most caudad intact
thoracic vertebra and the pelvis (Fig. 3). It has been noted that, in patients
with a myelomeningocele, the most caudad vertebra with radiographically
visible pedicles bilaterally usually correlates with the motor level1,10. We initially used this criterion
to classify our patients, but we found that it was not reliable.
However, we identified the most caudad vertebra with visible pedicles
for the purpose of discussing the outcomes of patients with different
motor and sensory levels.
A hip or knee contracture was graded as mild if it was £30°
in any plane, moderate if it was between 31° and 60°, and severe if
it was 61°. All but one of the foot deformities were rigid and could
not be corrected passively. The decision to perform an operation
on a lower extremity was based on the patient’s ambulatory
potential and on positional factors. Operative procedures to correct
deformities were done in patients who were able to walk or who were
believed to have the potential to do so11.
If a patient had a fixed lower-extremity deformity that interfered
with sitting or with wearing braces or shoes, an operation was done
to facilitate these functions. An attempt was made, either by closed
or open means, to reduce a dislocated hip in community ambulators.
Soft-tissue releases of hip and knee flexion contractures were done
to facilitate sitting by nonambulators and walking by those who
could walk. A supracondylar osteotomy of the distal part of the
femur was performed when severe flexion deformity of the knee was recalcitrant
to soft-tissue release. Operative procedures were performed on the
foot to provide a plantigrade surface and to allow shoe wear and
to avoid skin breakdown. In general, soft-tissue releases were performed
initially and osseous procedures were reserved for children in whom
soft-tissue procedures had failed.
In many patients, the motor examination did not reveal symmetrical
function. The results of the sensory examinations were more consistent,
and a dermatomal pattern could often be found. Many patients were
described as having "protective" sensation, which
meant that proprioception and sensation of light touch were present
but sensation could not be considered totally normal.
There were eight girls and five boys in Group I. The average age
was 1.9 years (range, one month to 5.5 years) at the time of the
first visit and 15.2 years (range, 6.2 to 29.7 years) at the time
of the last follow-up. The average duration of follow-up was 13.3
years (range, two to 26.6 years). Five children had a maternal history
of insulin-dependent diabetes mellitus, and one had a maternal history
of non-insulin-dependent diabetes. Seven patients had Type-A spinal
deformity, three had Type-B, and three had Type-C (see Appendix).
Group I, Type A (Seven Patients)
Twelfth thoracic level (one patient): Even though
the twelfth thoracic vertebra was the most caudad vertebra with
visible pedicles bilaterally in this patient (Case 1), anomalous
lumbar vertebrae with no clearly defined pedicles extended in the
midline to the fused ilia (Fig. 4). The motor power and sensation
in the lower extremities corresponded to the fifth lumbar level.
There was no scoliosis, and atlantoaxial instability due to an absent
odontoid process was treated with a posterior fusion from the occiput
to the third cervical vertebra. The patient was a community ambulator
with the aid of a knee-ankle-foot orthosis and Canadian crutches.
First lumbar level (one patient): Two remnants
of lumbar vertebrae with ill-defined pedicles articulated with the
pelvis in the midline in this patient (Case 2). Both lower extremities
were flail, and the sensory dermatomes were inconsistent, with normal
sensation in the right lower extremity and protective sensation
in the left lower extremity. A right congenital thoracolumbar scoliosis
from the second thoracic to the second lumbar vertebra measuring 85°
was treated with posterior spinal arthrodesis without instrumentation.
At the time of the last follow-up, the fusion was solid and the
curve remained at 85°. There were no cervical spine anomalies. The
patient was a community ambulator, with long leg braces on both
lower extremities, who walked with a swing-through gait.
Third lumbar level (three patients): In three
patients (Cases 5, 6, and 7), the most caudad vertebra with visible
pedicles was at the third lumbar level. All three patients had normal
sensation in the lower extremities, but the motor levels were less
consistent. One patient had motor power corresponding to the first
sacral level on one side and to the third lumbar level on the other
side, and one patient had motor power corresponding to the second
sacral level on one side and to the fourth lumbar level on the other
side. The third patient had motor power corresponding to the fourth
lumbar level bilaterally. No patient had scoliosis. One patient
had congenital fusion of the second and third cervical vertebrae, and
another had hypoplasia of the odontoid and the first cervical arch
and was treated with a posterior fusion from the occiput to the
atlas. All three patients were community ambulators, but two used
an ankle-foot orthosis and crutches to assist with walking.
Fourth lumbar level (two patients): In two patients
(Cases 3 and 4), the most caudad vertebra with visible pedicles
was at the fourth lumbar level. Both patients had motor power corresponding
to the first sacral level bilaterally. One patient had normal sensation
in both lower extremities, whereas the other had only protective
sensation. Neither patient had scoliosis or cervical spine anomalies. Both
patients were community ambulators, but one used a long leg brace
on the left to assist with walking.
All of the patients in Group I, Type A, were community ambulators.
These patients achieved this goal regardless of the motor levels
involved, and they often used braces and crutches for assistance.
At the time of writing, all of the patients continued to be community
ambulators.
Group I, Type B (Three Patients)
Twelfth thoracic level (one patient): This patient
(Case 10) had a block of anomalous lumbar vertebrae with no clearly
defined pedicles articulating with the right ilium away from the
midline. The patient had motor power corresponding to the second
lumbar level and sensation corresponding to the second sacral level
in both lower extremities. There was a right thoracolumbar congenital
scoliosis from the tenth thoracic to the second lumbar vertebra
measuring 62°. Bifid cervical vertebrae with congenital fusion of
the first three cervical vertebrae were present. The patient was
a household ambulator with bilateral long leg braces and Canadian
crutches. Neither the patient nor the family wanted operative treatment
of severe bilateral hip flexion contractures or the congenital scoliosis.
First lumbar level (two patients): In two patients
(Cases 8 and 9), the most caudad vertebra with visible pedicles
was at the first lumbar level. One patient had flail lower extremities,
whereas the other patient had motor power corresponding to the third
lumbar level bilaterally. The patient who had flail lower extremities
had sensation to touch, but not of pain, in the lower extremities,
and the other patient had normal sensation in the right lower extremity and
was insensate distal to the knee on the left. Neither patient had
scoliosis or cervical spine anomalies. Both patients were nonambulators
and used a wheelchair.
Only one of the three patients in Group I, Type B, was a household
ambulator, and this patient (Case 10) used long leg braces and Canadian
crutches following bilateral supracondylar osteotomy of the distal
part of the femur to correct knee flexion contractures.
Group I, Type C (Three Patients)
Eighth thoracic level (one patient): This patient
(Case 13) had flail and insensate lower extremities. The patient
had no scoliosis or cervical spine anomalies and was a nonambulator.
Twelfth thoracic level (two patients): In two
patients (Cases 11 and 12), the most caudad vertebra with clearly
visible pedicles was at the twelfth thoracic level. Both patients
had flail lower extremities with minimal protective sensation. One
patient had a right congenital scoliosis from the first to the eighth
thoracic vertebra measuring 25°, occipitalization of the atlas,
and fusion of the second and third cervical vertebrae. The other
patient had no scoliosis but had hemivertebrae at the second and
third cervical levels. Neither patient walked; both used a wheelchair.
All three patients in Group I, Type C, were nonambulators. In two
patients, the deformities at the hips and knees combined with thoracic
level motor function precluded walking, in spite of the deformities
at the hip being corrected in one patient (Case 12). The other patient
was not able to walk because of bilateral knee disarticulation performed
at another institution for the treatment of severe knee flexion
contractures. The patient had neither the strength nor the motivation
to use prostheses.
Group II
There were three girls and two boys in Group II. The average age
was 4.3 years (range, one to 10.2 years) at the time of the first
visit and twenty years (range, 16.1 to 24.7 years) at the time of
the last follow-up. The average duration of follow-up was 15.7 years
(range, 14.7 to 19.8 years). One child had a maternal history of
insulin-dependent diabetes mellitus, and four had no maternal history
of insulin-dependent or non-insulin-dependent diabetes. One patient
had Type-A deformity, two had Type-B, and two had Type-C.
Group II, Type A
Fourth lumbar level (one patient): This patient
(Case 14) had motor power corresponding to the second lumbar level
in both lower extremities, sensation corresponding to the first
sacral level in the right lower extremity, and sensation corresponding
to the fifth lumbar level in the left lower extremity. A left scoliosis
from the second thoracic to the third lumbar vertebra measuring
60° was treated with a combined anterior and posterior arthrodesis
without instrumentation. The curve measured 55° at the time of follow-up. There
were no cervical spine anomalies. The patient was a nonambulator.
Group II, Type B
Second lumbar level (two patients): There were
two patients (Cases 15 and 16) in whom the most caudad vertebra
with visible pedicles was at the second lumbar level. One patient
had motor power and sensation corresponding to the third lumbar
level bilaterally, and the other patient had motor power corresponding
to the second lumbar level but was insensate in the lower extremities.
Both patients had scoliosis. One had a posterior fusion with Harrington instrumentation
for the treatment of a congenital right curve from the first to
the tenth thoracic vertebra measuring 70°. At the time of follow-up,
the curve measured 53°. The other patient had a left curve from
the fourth to the ninth thoracic vertebra measuring 20° and a right
curve from the ninth thoracic to the first lumbar vertebra measuring
35°. Both curves were treated with observation. This patient also
had congenital fusion of multiple cervical vertebrae. Neither patient walked.
Group II, Type C
Twelfth thoracic level (two patients): In two
patients (Cases 17 and 18), the most caudad vertebra with visible
pedicles was at the twelfth thoracic level. One patient had flail
lower extremities, and the other patient had motor power corresponding
to the first lumbar level. Both patients had patchy sensation in
the feet. One patient had a 10° right scoliosis from the third to
the tenth thoracic vertebra and congenital fusion of the second
and third cervical vertebrae. The other patient had no scoliosis
but had occipitalization of the atlas with fusion of the second
and third cervical vertebrae. Neither patient walked.
Regardless of the type of deformity, no patient in Group II (associated
myelomeningocele) was able to walk, even when motor levels corresponded
to the third lumbar level, as they did in Case 15. All of these
patients used a wheelchair.
The etiology of lumbosacral agenesis is not fully understood. It
is believed to be caused by disruption in the development of the
caudad portion of the osseous spine and the spinal cord or by prenatal
exposure to various substances1,7,10.
The cause of this disruption in development is not known, and in
most reports no genetic predisposition has been noted4,7,10. No patient in the present series
had a relative with the same condition.
Lumbosacral agenesis is associated with maternal insulin-dependent
diabetes mellitus. In the present series, the mothers of six of
the eighteen patients took insulin during pregnancy. In the series
of Phillips et al.10, the mothers
of 50% of the patients had diabetes mellitus. Those authors
did not find a correlation between the severity of the deformity
and maternal diabetes. In one review, only three of 1150 children
born to diabetic mothers had sacral agenesis2.
We initially attempted to predict motor power and sensory levels
by identifying the most caudad vertebra with bilaterally visible
pedicles10. This vertebra was
not synonymous with the vertebra that articulated with the pelvis
or ilia. In other series, this landmark has been referred to as
the "nubbin," "last normal vertebra," "last
intact vertebra," or "last recognizable vertebra," with
no clear definition1,7,8. Phillips1 stated that the level of involvement
in these patients was best described by the most caudad normal vertebral
body, which "usually corresponds well with the degree of
motor impairment," but the term "normal" was
not defined. Phillips’s finding was reported by other authors
as well6,7, but our results did
not support this theory. There was only one patient (Case 16) in
whom the most caudad vertebra with bilaterally visible pedicles
corresponded with the motor power (the second lumbar level) and
no patient in whom the sensory level fully corresponded with this
vertebra. This finding is in contrast to observations in children
with only a myelomeningocele, in whom this landmark has been said
to be a more reliable predictor of motor power and sensory level1,10. Several patients had multiple
anomalous lumbar vertebrae that may have had intact neural elements
providing motor and/or sensory function (Fig. 4).
We did not find a correlation between the degree of hip dysplasia
and the level of spinal involvement or motor power. We believe that
hip reduction should be attempted only in patients who have ambulatory
potential. Of five hips treated with closed reduction (Cases 1,
2, and 3), four were maintained in the reduced position, and open
reduction was required in only one hip (Case 1, left) in our series.
Hip and knee flexion contractures should be corrected early in all
patients who have a potential to walk, and it is our impression
that deformities of all joints should be addressed during the same
operation.
Sixteen of the eighteen patients had knee deformities. Our findings
are in agreement with those of Phillips et al.10 and
Banta and Nichols7, who reported
that severe knee flexion contractures and popliteal webbing were
more common in patients with a high-level (first lumbar) lesion.
Banta and Nichols stated that knee flexion contractures need to
be corrected completely and that deformity can recur even with complete
correction. The majority of knee procedures in our series were done
on Group-II patients or those with severe knee flexion deformities
that precluded sitting.
All but one patient in the series had bilateral rigid foot deformities,
with clubfoot being the most common. Cast treatment was initially
attempted for most patients. In recalcitrant cases, if the patient
had ambulatory potential or if shoe wear was difficult, soft-tissue
releases were tried first. If the correction was inadequate, midfoot
or triple arthrodesis was then done. The feet were not treated in
only four patients, two with flail lower extremities and bilateral
clubfoot in Group I, Type C, and two (one with no foot deformity
and one with mild bilateral calcaneovalgus deformity) in Group II,
Type B.
We do not advocate amputation. Frantz and Aitken3 as well as Russell and Aitken5 believed that amputation allowed
the patient to sit better and to walk effectively with their hands.
Renshaw stated that amputation may be the treatment of choice for
patients with severe contractures of the lower extremities4. Winter noted that the bone harvested
from the amputation could be used for bone-grafting during spinal
reconstruction12. Andrish et al.6 and Banta and Nichols7 advocated leaving the lower limbs
alone in these patients since they provide stability for sitting,
often maintain proprioception and protective sensation, and provide
the patients with a better body image. In our series, the one patient
with bilateral knee disarticulation was not motivated to use prostheses. We
believe that soft-tissue releases performed early and selective
corrective osteotomies obviate the need for amputation in most patients.
In Group I, only one of the seven children with Type-A deformity
had scoliosis and one patient with Type-B deformity and one with
Type-C had a congenital thoracic or thoracolumbar curve. Four of
the five children in Group II had scoliosis, which was congenital
in one of them. In this small series of eighteen patients, we could
not correlate the presence of scoliosis with the pattern of deformity
or motor level, other than to observe that it was more prevalent
in children with a myelomeningocele (Group II). The curves in these
patients should be observed and arthrodesis should be considered
for progressive curves or curves that interfere with walking or
sitting.
Nine of the eighteen patients had anomalies of the cervical spine,
especially in the cephalad region. Three of the seven patients in
Group I, Type A, had cervical spine anomalies. Seven patients in
the series had various combinations of cervical hemivertebrae and
congenital fusion without instability, and two others were treated
with a posterior arthrodesis for odontoid hypoplasia and atlantoaxial
instability. We believe that the cervical spine should be examined
radiographically for possible anomalies or instability in all patients
with lumbosacral agenesis.
Our patients functioned well without a spinopelvic fusion, a procedure
advocated by Perry et al.8. Some
authors believed that spinopelvic fusion allows the hands to be
free from supporting the body, the viscera to be protected from
compression, and lower extremity contractures to be stretched more
effectively. This reasoning has been used recently by several authors2,13,14 who have described modified
techniques with modern instrumentation. Even though we have no experience
with spinopelvic fusion, we believe that patients who have hip flexion contractures
will have difficulty sitting and walking if the spine is fused to
the pelvis. Andrish et al.6 reported
on nine patients with spinopelvic instability who had no visceral
compression and who were able to walk better because they had a
mobile spinopelvic junction. Banta and Nichols7 believed
that lumbopelvic instability did not need to be treated operatively
and that a brace could be used if support was necessary.
In Group I, patients with Type-A deformity—that is,
with the spine in the midline—did better in terms of walking
than did those with Type-B or Type-C deformity. All seven patients with
Type A were community ambulators, most with the aid of braces and
crutches. It is for this reason that lower-extremity deformities
in these children should be corrected early, to facilitate walking.
Of the three patients with Type-B deformity, only one (Case 10)
was a household ambulator. This patient probably would have walked
better if the hip flexion contractures had been corrected. Group-I,
Type-A patients have a potential to walk at least in the house;
thus, their ambulatory potential should be assessed early in childhood,
and all lower-extremity deformities should be corrected. None of
the patients in the other subgroups were community ambulators, and
extensive operations for correction of deformities are not needed
for such patients except to aid in sitting or wearing of normal
shoes.
In conclusion, we propose a new classification for patients with
partial or total agenesis of the lumbar spine and complete agenesis
of the sacrum. Using our classification, we found that patients
in whom the vertebral column articulates with the pelvis in the
midline and who do not have a myelomeningocele have the best potential
for walking. No patient with a concomitant myelomeningocele was
able to walk. Identifying the most caudad vertebra with bilaterally
visible pedicles was of no predictive value in determining motor
or sensory levels or ambulatory potential. Radiographs of the cervical
spine should be reviewed for congenital anomalies or instability,
as 50% of our patients had these problems. The spine should
be fused if the patient has a scoliosis that is progressive or interferes
with sitting or walking. Selective operative releases and osseous
procedures on the hip, knee, and foot should be reserved for patients
with ambulatory potential and to facilitate sitting or shoe or brace
wear. We do not advocate limb ablation or spinopelvic fusion. Using
this new classification, we can select patients who have ambulatory
potential and thus are candidates for corrective surgery of the
lower extremities that will enhance walking.