The three described clinical entities of Langerhans cell histiocytosis
(formerly known as histiocytosis X) are Letterer-Siwe disease,
Hand-Schüller-Christian disease, and eosinophilic granuloma. Once thought
to be three separate entities, these disorders are now recognized as different
manifestations of the same disease process—a clonal proliferation of
Langerhans
cells1,2.
Langerhans histiocytes are dendritic cells whose main role is in presenting
antigens. They are commonly found in isolation in the epidermis; however, they
are also normally present in small numbers throughout the body. Letterer-Siwe
disease, the most malignant form, leads to widely disseminated clusters of
Langerhans cells throughout the body. Hand-Schüller-Christian disease
classically presents with a triad of a skull lesion, exophthalmos, and
diabetes insipidus. Eosinophilic granuloma is a benign proliferation of
Langerhans histiocytes occurring in a unifocal or multifocal manner that
commonly affects the skeletal
system3.
In a large series of 314 patients with Langerhans cell histiocytosis seen
at the Mayo Clinic, 188 (60%) had at least one skeletal
lesion4. The most
frequent sites of these skeletal lesions were, in descending order, the skull,
femur, mandible, pelvis, and spine. Despite heterogeneous treatment, the vast
majority of the patients did well, and there was no evidence suggesting that
any treatment was more advantageous than
another5. A recent
study by Ghanem et al. also indicated that children with only skeletal lesions
have excellent outcomes, whereas those with systemic involvement have a widely
variable potential for
healing6.
Combining the data from the large series of children with skeletal
Langerhans cell histiocytosis seen at the Mayo Clinic with the data from two
other large series (forty and fifty-three children) showed that forty-six
(17%) of 265 children with skeletal lesions had spinal
involvement1,5,7.
These series suggest that lesions in the spine are a unique feature of the
pediatric population with Langerhans cell histiocytosis, in contrast with the
adult population with the disorder. As a result of the low frequency with
which vertebral lesions are encountered in clinical practice, there are very
few studies in the literature focusing on the spine. We found several series
of ten or fewer
patients8-12,
but only four that included more than ten
patients13-16.
We analyzed the distribution and presentation of spinal Langerhans cell
histiocytosis in twenty-six children diagnosed at our institution between 1970
and 2003. Data on treatment and long-term outcomes are presented for
twenty-three children in this group who were followed for at least two years.
To the best of our knowledge, this is the largest series of spinal lesions
reported to date. The objective of this study was to evaluate the distribution
and extent of disease at presentation as well as to determine the long-term
prognosis of the disease in terms of both disease recurrence and progressive
spinal deformity or instability. We hypothesized that spinal deformity is more
likely to develop in patients who present with asymmetric vertebral collapse
than it is in patients who present with symmetric collapse.
Twenty-six children with a biopsy-proven diagnosis of Langerhans cell
histiocytosis and at least one lesion in the spinal column were treated in our
department between 1970 and 2003. The group included fourteen boys and twelve
girls with a total of forty-four involved vertebrae. The mean age at the time
of the diagnosis was 8.2 years (range, 0.2 to 16.4 years). A biopsy of a
spinal lesion established the diagnosis in sixteen patients, whereas a biopsy
of a more easily accessible location in the skeleton established the diagnosis
in the remaining ten patients.
The clinical presentation, radiographic presentation, operative records,
treatment, findings at the follow-up evaluations, and follow-up radiographs
were reviewed. Twenty-three patients were followed clinically for two years or
more (Table I). Two others had
been diagnosed less than two years before the time of the review, and one had
been lost to follow-up. The mean duration of follow-up for the twenty-three
patients was 9.4 years (range, two to 22.7 years). All twenty-six patients are
described in the sections on clinical and radiographic presentation, but only
the twenty-three patients who had been followed for at least two years are
included in the sections on treatment and long-term outcomes.
Investigators at our institution reviewed the diagnostic radiographs of
twenty-four of the twenty-six patients. The radiographs of the other two
patients were not available, and therefore the radiology reports were used to
define the extent of the vertebral collapse. Two of the authors of this study
(S.G. and J.P.D.) performed the radiographic grading, and they agreed on all
classifications. To test our hypothesis, a new classification system for
vertebral collapse based on both morphology and the extent of maximal collapse
was created. Vertebral body collapse was classified as grade I (0% to 50%
collapse) or grade II (51% to 100% collapse) and was subclassified as either A
(symmetric collapse) or B (asymmetric collapse, such as lateral, anterior, or
posterior wedging). Lesions of the posterior elements of the spine (transverse
process, spinous process, facet joints, pedicle, and/or lamina) were
classified as grade III (Fig.
1). The classic finding of vertebra plana would be assigned a
grade of IIA with this system. The extent of maximal collapse was assessed by
measuring the vertebral height and comparing it with the mean of the heights
of the vertebrae immediately cephalad and caudad to the affected vertebra.
Statistical analysis was done with use of JMP IN statistical software
(version 4.04; SAS Institute, Cary, North Carolina, 2001). Probability testing
for the distribution of the lesions was done with use of the null hypothesis
that there is a 7 of 24 probability of a lesion being found in the cervical
spine, a 12 of 24 probability of a lesion being found in the thoracic spine,
and a 5 of 24 probability of a lesion being found in the lumbar spine. The
Pearson chi-square test was used to determine significance in all frequency
testing.
Distribution
The forty-four vertebral lesions found in our twenty-six patients were
distributed throughout the spinal column. Twenty lesions (45%) were found in
the cervical spine; fourteen (32%), in the thoracic spine; and ten (23%), in
the lumbar spine. The twenty cervical lesions were found in a total of eleven
patients (six male and five female); the fourteen thoracic lesions, in ten
patients (six male and four female); and the ten lumbar lesions, in ten
patients (six male and four female). Two (5%) of the forty-four lesions were
located in the posterior elements of the vertebra, whereas the other forty-two
lesions occurred in the vertebral body. Both of the posterior lesions were in
the cervical spine. This distribution revealed a predilection for the cervical
spine (p = 0.02), and there were fewer lesions in the thoracic spine than
had been expected given its large number of segments relative to the entire
spinal column (p = 0.006). With the numbers available, neither gender nor
age at the time of the diagnosis had a significant relationship with the
distribution of lesions (p > 0.5 for both).
Of the twenty-six patients, fifteen had involvement of a single vertebra
and eleven had involvement of multiple vertebrae. Multiple vertebral
involvement was found in seven of the eleven children with cervical lesions
and five of the ten children with lumbar lesions. Both cervical disease (p
= 0.03) and lumbar disease (p = 0.04) were associated with multi-level
involvement.
Eleven of the twenty-six patients had evidence of disease in the skeletal
system outside the vertebral column. There were six lesions in the femur; five
lesions in the humerus, ribs, and skull; four lesions in the pelvis; two
lesions in the tibia, clavicle, and mandible; and one lesion each in the
patella, scapula, and sternum. These lesions were discovered with diagnostic
radiographic studies such as skeletal surveys and technetium bone scans. At
least one of those studies was performed in twenty-four of the twenty-six
patients. The two patients without secondary imaging studies may have had
additional, nonsymptomatic lesions; however, this cannot be determined
retrospectively in the absence of those studies. Of the eleven patients with
extraspinal disease, five had involvement of only one vertebra whereas six had
involvement of multiple vertebrae. Overall, sixteen (62%) of the twenty-six
patients had more than one skeletal lesion. Only two (8%) of the twenty-six
patients were found to have extraskeletal disease. The extraskeletal lesions
were located in the lungs and bone marrow in one patient and in the lungs and
liver in the other. Both of those patients also had involvement of multiple
vertebrae and of skeletal sites outside the spinal column.
Clinical and Radiographic Presentation
Back or neck pain was the presenting symptom in all twenty-six patients,
and it was the only presenting symptom in seventeen. Four patients had the
additional finding of torticollis, and three had abnormal gait. Only three
patients presented with neurologic symptoms, and none had a neurologic
deficit. The symptom in each case was pain radiating down the upper extremity
as a result of a cervical lesion. The neurologic symptoms, the torticollis,
and the gait abnormalities all resolved with treatment of the underlying
disease.
Thirty of the forty-four involved vertebrae had symmetric collapse, and
twelve had asymmetric collapse. The radiographic classification was grade IA
for twenty vertebrae, IB for three, IIA for ten, IIB for nine, and III for
two. Grade-I lesions were significantly associated with symmetric collapse (p
= 0.03), whereas grade-II lesions were not associated with either symmetric
or asymmetric collapse.
Treatment
Treatment of the twenty-three children who had been followed for two years
or more consisted of chemotherapy (combinations of oral methotrexate, oral
prednisone, or intravenous vinblastine) in ten patients and radiation therapy
in four patients. Two patients had both chemotherapy and radiation therapy.
All four patients who received radiation therapy were treated prior to 1991.
Six of the ten patients treated with chemotherapy had multiple vertebral
lesions and seven of the ten had extraspinal lesions. All but one of the
patients who received radiation therapy had both involvement of multiple
vertebrae and extraspinal lesions. Of the two patients with extraskeletal
disease, both received chemotherapy and one also received radiation therapy.
All other patients were treated symptomatically and followed clinically and
radiographically. A spinal orthosis was used only for comfort for a few weeks
following the biopsy.
Clinical Follow-up
Of the twenty-six patients, twenty-three with a total of forty-one involved
vertebrae were followed clinically for at least two years (mean, 9.4 years;
range, two to 22.7 years). None of the twenty-three patients with adequate
clinical follow-up had evidence of persistent or active disease at their last
clinical encounter. All presenting signs and symptoms of the disease had
resolved. None of the patients who had received radiation therapy had a
secondary malignant lesion at the time of the most recent follow-up.
Some degree of spinal deformity developed in four patients with long-term
follow-up (Table II). Spinal
deformity was defined as scoliosis of >10° or a sagittal plane
deformity that was >10° greater than the normal range of cervical
lordosis, thoracic kyphosis, or lumbar lordosis. Two of these patients had a
grade-IIA lesion, and two had a grade-IIB lesion. On the basis of these
results, the hypothesis that asymmetric collapse predisposes patients to the
development of spinal deformity was rejected. In our series, both major and
minor deformities developed both early and late after diagnosis. There was no
relationship between the development of deformity and the initial grade of the
vertebral collapse.
Clinical and Radiographic Presentation
Children with Langerhans cell histiocytosis of the spine most often present
with dull back pain as their only
symptom8-15,17.
This was true for our series, with all patients complaining of back or neck
pain at the time of presentation. Neurologic symptoms were very uncommon,
occurring in only three patients. The three patients all had shooting pain in
the upper extremity that resolved promptly following the biopsy. Two of the
three patients had chemotherapy with oral prednisone and methotrexate. Our
findings support prior descriptions of prompt resolution of neurologic
symptoms following treatment of spinal lesions in earlier case
reports8,18-20.
Interestingly, the classic radiographic presentation of spinal lesions as
vertebra plana was the exception rather than the rule. In fact, although
severe (grade-II) lesions were equally divided between patients with symmetric
and asymmetric collapse, less severe (grade-I) lesions were significantly
associated with symmetric collapse (p = 0.03).
Diagnosis
Some authors have recommended not performing a vertebral biopsy if the
lesion has the classic radiographic characteristics of Langerhans cell
histiocytosis and can be followed
closely16,21.
The classic radiographic findings include vertebral collapse, maintenance of
disc spaces, lack of extraspinal spread, and lack of a soft-tissue mass.
Magnetic resonance imaging is performed to rule out a soft-tissue mass, which
would suggest a more aggressive cause of vertebral collapse. The advantages of
this approach are avoidance of an unnecessary biopsy in typical cases and
avoidance of the very low risk of inadvertently damaging the vertebral growth
plates during
biopsy7,21.
On the basis of the findings in this study and the excellent long-term natural
history observed in this study, this strategy for deciding when to perform a
biopsy seems prudent.
We believe that a tissue diagnosis is essential in suspected cases of
spinal Langerhans cell histiocytosis only if there are any atypical features
such as a soft-tissue mass, disc space involvement, or neurologic symptoms
(Figs. 2-A, 2-B, and
2-C). Children with typical
radiographic findings of spinal Langerhans histiocytosis should be followed
closely clinically and radiographically to ensure that the lesion is benign
and resolving. Children presenting with vertebra plana can potentially have a
variety of different malignant tumors, including Ewing sarcoma, osteosarcoma,
leukemia, and lymphoma, which underscores the need for an appropriate and
thorough work-up (including biopsy when appropriate) and good
follow-up22-25.
If multiple skeletal sites are involved, only the most accessible site needs
to be biopsied to confirm the diagnosis of Langerhans cell
histiocytosis14,15.
Distribution of Disease
Only two (8%) of the twenty-six patients in our study had extraskeletal
disease. This finding is consistent with those of four large previous studies,
in which a total of six (10%) of sixty-two patients with Langerhans cell
histiocytosis involving the spine also had extraskeletal
disease13-16.
The predominance of isolated skeletal disease may reflect an actual trend in
patients with lesions involving the spine, but it may also be due to referral
bias at our institution. It is worth noting that, in the large study of 314
patients seen at the Mayo
Clinic4, spanning
nearly fifty years, fewer spinal lesions were identified than were seen in the
current study. Moreover, since our cases were collected with use of our
institution's comprehensive musculoskeletal tumor database, we believe that
our series included all children with spinal lesions seen at our
institution.
In our series, twenty (45%) of the forty-four lesions were found in the
cervical spine, a much higher percentage than described in the four largest
previous case
series13-16,
in which a total of twenty-two (19%) of 113 lesions were in the cervical
spine. This difference may be a reflection of the interest of the senior
author (J.P.D.) in the pediatric cervical spine; however, to our knowledge,
all patients who presented to our institution with spinal lesions were seen in
our department. In contrast, there were relatively fewer lesions in the
thoracic spine than we had expected, given the large size of this segment as
compared with the cervical and lumbar segments. Only fourteen (32%) of the
forty-four lesions were in the thoracic region, as opposed to fifty-five (49%)
of 113 lesions in the four largest previous case series. The skewed frequency
distribution in our series was strongly significant for the cervical spine
predilection (p = 0.02) and for the relative infrequency of thoracic
lesions given the large size of the thoracic spine (p = 0.006). Moreover,
cervical and lumbar lesions were found to be significantly associated with
multilevel disease (p = 0.03 and p = 0.04, respectively). Neither of
these novel findings was identified in the previous large studies of spinal
lesions13-16.
Because of the high frequency of multiple skeletal lesions in patients who
are found to have a spinal lesion, we recommend that a technetium bone scan or
a skeletal survey be performed early in the evaluation of every child with a
suspected spinal lesion. Although all of the initial spinal lesions in the
children in our series were diagnosed primarily after the patient complained
of back pain, all additional lesions were found on secondary imaging. The
finding of multiple sites of disease allows the physician to have increased
confidence in the diagnosis since other entities that may present with
vertebral collapse do not occur at multiple sites. Furthermore, this
information can help to determine whether chemotherapy is needed. Finally, it
is important to monitor all sites of disease for progression and pathologic
fracture.
The decision whether to perform a technetium bone scan or skeletal survey
in the initial phases of the diagnosis continues to be debated in the
literature26-28,
and there is no conclusive evidence showing the superiority of one study over
the other. The fact that the total radiation dose is less in a technetium bone
scan than it is in a skeletal survey led us to prefer the first
method28,29.
A large, multicenter, prospective study is needed to identify the best
screening tool.
Treatment
The current study covers a period of thirty-three years, and a variety of
treatment modalities were used. Almost all of the patients in our series who
were treated during the 1970s and 1980s were given chemotherapy or radiation
therapy, or both. Oral chemotherapy with prednisone or methotrexate, or both,
continues to be commonly used for multifocal skeletal and systemic Langerhans
cell
histiocytosis4,30-32.
Patients who did not receive chemotherapy or radiation therapy were followed
clinically and with radiographs. Spinal orthoses were not used for treatment
of the spinal lesions; rather, they were worn, for comfort, for only a few
weeks following the biopsy.
In the past, low-dose radiation therapy has been advocated for spinal
lesions that continue to cause pain following
biopsy1,33-35.
It should be noted, however, that secondary malignant lesions following
radiation therapy for systemic Langerhans cell histiocytosis were reported in
four (4%) of eighty-nine
patients36. Of
these four patients, two who were two and eighteen years of age died of
leukemia five and seventeen years, respectively, following the radiation
therapy. The other two patients, who were two months and three years of age,
died of thyroid cancer at twenty-eight and fifteen years after
treatment36.
Radiation therapy also poses a risk to the vertebral end plates, which are
usually left unaffected by the disease process itself. Although we are not
aware of any reported cases of growth arrest or secondary sarcoma following
radiation therapy for spinal lesions, we believe that the risk associated with
radiation therapy is unacceptable given the overwhelmingly positive natural
history of this disorder. Radiation therapy is no longer used for this
indication at our institution. This view is in agreement with those set forth
in other recent large case
series7,14,37.
Except for open biopsy of the affected vertebra, surgery is rarely
indicated for the spinal lesions in these patients. Normally, the only
indications for spinal surgery are stabilization of an unstable segment of the
spine that cannot be stabilized with an orthosis or neurologic symptoms due to
compression of the spinal cord by the collapsed vertebra. There have been
sporadic case reports of spinal lesions requiring surgical
stabilization7,9,38.
In most patients, any spinal instability present following biopsy and
curettage of the lesion can be supported in an orthosis until sufficient
reconstitution has occurred to restore stability.
Long-Term Follow-up
Spinal Deformity
Initially, we hypothesized that a tilted vertebral body would be more
likely to lead to scoliosis or kyphosis than would a flat vertebral body. We
believed that the distorted shape had implications with regard to spinal
balance, particularly in the growing spine. Our results failed to support this
hypothesis. Both children who required surgery for spinal deformity had a
grade-IIA lesion, and the other two children with spinal deformity had a
grade-IIB lesion (Table III).
Significant deformity occurred both early (Case 17) and late (Case 23) after
the diagnosis (Figs. 3-A, 3-B,
and 3-C). Although our initial
hypothesis was rejected, since deformity occurred following both symmetric and
asymmetric collapses, the results in this series underscore the need for close
follow-up, through skeletal maturity, of all children with spinal lesions. We
recommend radiographs every three months for the first year after the
diagnosis to confirm the benign nature of the disease and annual radiographs
thereafter, through skeletal maturity, to monitor spinal growth and balance.
Any child with multifocal disease should be referred to an oncologist so that
a decision can be made regarding the need for chemotherapy.
Natural History of the Disease
Our patients had excellent long-term results, a finding that is consistent
with those in smaller previous studies of spinal
lesions8-16.
Despite differences in the types of treatment, which included chemotherapy,
radiation therapy, both chemotherapy and radiation therapy, and no treatment,
none of the twenty-three patients who were followed clinically for at least
two years had any clinical evidence of disease. Use of a spinal orthosis does
not appear to be necessary given the excellent results in this study without
uniform use of orthoses. Aside from the two patients who had spinal fusion,
the remaining patients are fully active children and young adults with no
limitations of their activity or range of motion of the spine. The results had
no relationship to the type of therapy that was used, a finding that adds to
the body of evidence indicating that the natural history of these spinal
lesions is that they resolve on their own.
There have been only a few studies with sufficient radiographic evidence to
quantify the extent of reconstitution, and all showed excellent results,
particularly in younger
children13,15.
Many of our patients had excellent reconstitution of vertebral height
following diagnosis and treatment (Figs.
4-A, 4-B, and
4-C). Unfortunately, we did not
have enough radiographic data to draw any substantial conclusions regarding
long-term reconstitution of vertebral height.
Note: The authors acknowledge Rajesh V. Patel, MD, for
assistance in the initial review of radiographs.