After institutional review board approval had been obtained, a
retrospective review of our institution's Pediatric Tumor Registry from 1970
to 2003 was conducted to identify children with non-Hodgkin's lymphoma in
bone. Over this period, 306 cases of non-Hodgkin's lymphoma in children were
identified. In eighteen of these cases, the primary site of involvement was
bone.
The present study included patients with non-Hodgkin's lymphoma that was
limited to one or more osseous sites, including patients with bone-marrow
involvement, provided that they had no evidence of other systemic involvement.
Patients were excluded if they had a diagnosis other than non-Hodgkin's
lymphoma, if they had evidence of systemic involvement other than bone-marrow
involvement, if they had non-Hodgkin's lymphoma with secondary involvement of
bone, and if they had less than two years of follow-up. Of the eighteen cases
that were identified, fifteen were included in our review. Two patients were
excluded because they were seen at our institution for a second opinion only
and therefore had insufficient data. One patient was excluded because there
had been less than two years of follow-up. The individual characteristics of
the fifteen patients in the study group are presented in the Appendix. The
study group included ten male and five female patients with a mean age of 11.6
years (median, 12.1 years; range, 1.2 to 19.1 years) at the time of
presentation.
Clinical information was obtained from patient medical records.
Retrospective evaluation included collection of demographic data, clinical and
radiographic information, pathology reports, treatment information, and
follow-up evaluations. All fifteen patients were included in the survival
analysis. Ten patients survived, with a mean duration of followup of 13.6
years (median, 17.2 years; range, two to twenty-three years), and five
patients died, with a mean interval from presentation to death of 2.1 years
(median, 1.9 years; range, 0.4 to 4.9 years).
At the time of presentation, a complete evaluation was performed to rule
out systemic lymphoma. The clinical staging methods differed among patients
but variably included routine serum chemistry studies; a complete blood-cell
count; urinalysis; lumbar puncture; posteroanterior and lateral chest
radiographs; computed tomography of the chest, abdomen, and pelvis; an
intravenous pyelogram; gallium scanning; abdominal and pelvic ultrasound;
magnetic resonance imaging; and a technetium bone scan. All patients had plain
radiographic imaging of the primary site. All patients had a bone-marrow
examination (involving aspiration, biopsy, or both).
Considerable inconsistencies were found in the literature regarding the
classification of primary lymphoma of bone in children. We therefore created a
classification system in which patients were described as having localized
disease (solitary-bone involvement, no marrow involvement), diffuse disease
(multiple-bone involvement, no marrow involvement), or disseminated disease
(bone and marrow involvement, no other systemic involvement).
Pathology slides of twelve of the fifteen cases were available for
histologic review and were reexamined by a single hematopathologist (J.K.C.)
at our institution. They were assigned revised World Health Organization
classifications (see
Appendix)16.
Pathology reports, including the current World Health Organization
classification, were available in the three cases in which tissue was not
available for review. Lymphoblastic markers (TdT, MIC2, O13) were used to
classify the subtypes. Staining for immunologic markers of B cells (CD20,
CD79a, Pax-5) and T cells (CD3) were also used to further classify the tumor
immunophenotypes by defining the types of malignant cells (B or T cells) that
were present in the lesions.
A survival analysis for the patients in the present series was conducted to
test the association between certain demographic, clinical, radiographic, and
pathologic characteristics and survival. The primary end points for this
analysis included relapse (defined as recurrent disease after treatment,
either at the primary site or at a metastatic site) and death. The factors
that were analyzed included gender, age, the type of bone involved (i.e., long
bones, flat bones, vertebral bones), multiple-bone involvement, the presence
of constitutional symptoms (including weight loss of 10% of baseline over six
months, a fever of 38°C, and/or drenching night sweats), swelling or a
mass on physical examination, tenderness to palpation, histologic subtype, a
positive result on initial bone-marrow examination, a positive result on
repeat bone-marrow examination, radiographic evidence of soft-tissue extension
or pathologic fracture, anemia (defined as a hemoglobin level of <13 g/dL
in male patients and <11.5 g/dL in female patients), elevated platelet
count, elevated serum calcium (>10.2 mg/dL), elevated lactate
dehydrogenase, elevated alkaline phosphatase, treatment regimen (chemotherapy
or chemotherapy combined with radiation therapy), and treatment toxicity
resulting in a change in regimen (defined as toxicity necessitating a
reduction in the dose of chemotherapy or radiation, a change in agent, or
discontinuance of the original regimen).
Using the same methodology that was applied to our patients, we considered
prognostic factors in pooled data from all of the case series reported in the
literature, including the present data, that have exclusively analyzed primary
lymphoma of bone in
children1-4,6,7.
We did not include patients who were found as part of a larger adult series.
Although a total of 122 total patients were available when our patients were
combined with those from the literature, the survival analysis did not include
thirty-one children from one study of localized primary lymphoma of bone
because accurate data for specific patients could not be determined from that
report5. A modified
Murphy staging system was used to classify the patients in the literature as
stage I (IE) (involvement of a single osseous site; i.e., localized disease),
stage II (involvement of a single osseous site and regional lymph nodes; i.e.,
localized disease), stage III (involvement of multiple osseous sites without
marrow or node involvement; i.e., diffuse disease), and stage IV (metastatic
disease including bone-marrow involvement; i.e., disseminated
disease)3. The
factors that were analyzed for the ninety-one patients pooled from other
studies in the literature with adequate data included age, gender, the type of
bone involved (i.e., long bones, flat bones, vertebral bones), multiple-bone
involvement, the degree of involvement (stage), and the type of treatment
(chemotherapy alone or combination therapy). These factors were evaluated with
respect to survival.
Statistical Methods
Statistical analysis was done with use of S-PLUS statistical software. A
product-limit univariate analysis (Kaplan-Meier analysis) was performed to
analyze the prognostic significance of various factors. Probability testing
was conducted against the null hypothesis that there was no difference in
survival between patients with a tested characteristic and those without it. A
standard log-rank test was used to test for differences in the time to an
event (death or relapse), and the level of significance was set at p <
0.05. Ninety-five percent confidence intervals were determined by a normal
approximation using 1.96 times the standard deviation of each point estimate.
Although the confidence intervals of the two data curves may have overlapped
at individual data points, significance was determined with use of the
log-rank test and the summation of the difference across the entire curve.
Clinical Presentation
The clinical presentation of the individual patients is presented in a
table in the Appendix. The most common presenting complaint was pain without
antecedent trauma, which occurred in twelve of the fifteen patients. Other
symptoms included swelling or mass (five patients), fever (four patients),
weight loss (two patients), night pain (two patients), limp (three patients),
and irritability (one patient). One patient presented with a pathologic
fracture of the distal part of the femur after a fall. Another patient
presented with neurologic symptoms consistent with a T9 sensory level,
referable to a lesion in the T6-T9 vertebral bodies. Constitutional symptoms
were present in four patients. Two patients had a generally "ill"
appearance; one child appeared cachectic, and the other child presented with
decreased responsiveness and hypotonia. The physical examination revealed
swelling or a mass and tenderness to palpation in five patients, only swelling
or a mass in three patients, and only tenderness to palpation in two patients.
In five patients, neither tenderness nor a palpable mass was detectable on
physical examination. There was no relevant family history in any of the
patients studied.
The mean delay from the onset of symptoms until the final diagnosis was 6.2
months (range, 0 to 2.5 years). Twelve of the fifteen patients had a delay of
more than one month and eight had a delay of more than three months. The
causes of delay were multiple, and it is impossible to separate the individual
contributors. Most often, this was due to a delay in the initial presentation,
with patients describing a prolonged period of nonspecific pain and/or
swelling before presenting to a physician. Occasionally, the physician
attributed these nonspecific symptoms to other causes of musculoskeletal pain,
such as muscle strain or synovitis. In several cases, the correct diagnosis
was delayed because the histological findings were difficult to interpret. In
five patients, the initial pathology report was not interpreted as
demonstrating lymphoma and was misinterpreted in one patient each as Hodgkin's
disease or eosinophilic granuloma, osteosarcoma, histiocytosis X, chronic
sterile osteomyelitis, and histiocyte proliferation versus osteomyelitis.
Site of Involvement
The distribution of the specific bones involved is illustrated in
Figure 1. A total of forty-six
osseous lesions were observed in the fifteen patients studied, and the most
common sites of involvement were the pelvis (eleven), the femur (nine), and
the spine (eight). Eight patients had single-bone involvement, and seven had
multiple-bone involvement. Overall, the mean number of bones involved was 3.1
per patient (range, one to twelve bones per patient).
Laboratory Values
The laboratory studies that had been performed at the time of presentation
were examined. Abnormal findings included anemia (eight patients), elevated
erythrocyte sedimentation rate (six patients), elevated platelet count (five
patients), elevated lactate dehydrogenase (five patients), elevated serum
calcium (three patients), and elevated alkaline phosphatase (three patients).
None of the patients studied had an elevation of the white blood-cell
count.
Extent of Disease
The extent of disease for individual patients is listed in the Appendix.
Seven patients had localized disease, three had diffuse disease, and five had
evidence of bone-marrow involvement (disseminated disease). Two of the latter
five patients had an initially negative bone-marrow examination and then had a
repeat bone-marrow biopsy that demonstrated tumor involvement. Four of the
five patients with disseminated disease had pelvic bone involvement, and four
of these patients had multiple-bone involvement. Of the seven patients with
multiple-bone involvement, four patients had a positive bone-marrow
examination, whereas only one of the eight patients with solitary-bone
involvement had evidence of disseminated disease.
Radiographic Imaging of the Primary Site
All fifteen patients had a plain radiographic examination. Forty-six
lesions were identified. Plain radiographs were available for twenty-three
individual lesions, including all of the primary lesions as well as some of
the secondary lesions. Plain radiographs of the remaining lesions were not
available because these sites were studied only with other radiographic
modalities. Two patients had radiographic findings that were interpreted as
normal at the time of initial presentation. On plain radiographs, the majority
of lesions demonstrated radiolucency (sixteen of twenty-three lesions) and/or
sclerosis (ten of twenty-three lesions). Nine lesions demonstrated both
radiolucency and sclerosis, seven lesions demonstrated only radiolucency, one
lesion demonstrated only sclerosis, and six lesions demonstrated neither
sclerosis nor radiolucency. The radiographs of these twenty-three lesions
demonstrated evidence of pathologic fracture in association with four lesions,
periosteal response in association with seven lesions, and soft-tissue mass
association with three lesions.
Additional imaging studies of the primary lesion were variably obtained.
Computed tomography of the primary lesion was performed for eight patients.
Five of these patients had lesions that appeared radiolucent, permeative, or
destructive, and two of these lesions also had evidence of sclerosis. Two
lesions demonstrated abnormal marrow characteristics, and one lesion was
characterized only by sclerosis. Three studies demonstrated evidence of a
soft-tissue mass. Both patients who had negative plain radiographs had a
positive computed tomographic scan. Magnetic resonance imaging was performed
for six patients. Of these, four patients had had previous computed tomography
whereas two had had only magnetic resonance imaging to further evaluate the
lesion. Lesions were characterized by low signal intensity on T1-weighted
images and high signal intensity on T2-weighted images (Figs.
2-A through 2-D). Magnetic
resonance imaging demonstrated a soft-tissue mass in two patients in whom
computed tomography had failed to demonstrate soft-tissue involvement.
Overall, six of the fifteen patients had a soft-tissue mass on radiographic
studies, with three soft-tissue masses demonstrated only with additional
studies.
Histologic and Immunologic Features
The histologic subtypes of the individual patients are listed in the
Appendix. According to the revised World Health Organization classifications,
the majority of cases were classified as diffuse large-cell lymphoma (nine
patients) or lymphoblastic lymphoma (four patients). One patient had Burkitt
lymphoma. In one patient, the histologic diagnosis was unclear and the lesion
was described as a CD30+ lymphoma. Although demonstrating some features
consistent with Hodgkin's disease, the specimen had characteristics of
anaplastic large-cell lymphoma. Because of this finding, the patient was not
excluded from the results. The immunophenotype (i.e., determination of the
malignant cell of origin), available for thirteen patients, was classified as
B cell in nine patients and pre-B cell in four.
Therapy
Two patients had a surgical procedure other than a biopsy. One patient had
an emergent thoracic laminectomy because of spinal cord compression, and the
other a multiple-rib resection with removal of associated soft tissues at the
time of diagnosis prior to the initiation of chemotherapy. All fifteen
patients received chemotherapy. Nine patients received chemotherapy alone,
whereas six received a combination of chemotherapy and radiation therapy. The
patients who received radiation therapy were diagnosed earlier in our study,
when radiation was used more frequently as an adjunct to chemotherapy for
lymphomas.
Thirteen patients had documented toxicity to the treatment and, as a result
of this toxicity, seven patients required a change or discontinuance of the
treatment regimen. Toxicity included neurologic abnormalities such as ptosis,
tingling of the peripheral extremities, obstipation, vocal cord paralysis, or
foot drop (in patients managed with vincristine); hemorrhagic cystitis (in
patients managed with cyclophosphamide); renal toxicity with a decrease in
glomerular filtration rate; allergy; cholestatic jaundice (in patients managed
with 6-mercaptopurine and methotrexate); and pancreatitis (in patients managed
with L-asparaginase). There were no reported complications in the patients
receiving radiation therapy.
Long-Term Follow-up
The outcomes for the fifteen patients for whom follow-up was available are
presented in the Appendix. The mean duration of follow-up for the ten patients
who survived was 13.6 years (median, 17.2 years; range, two to twenty-three
years). All of these patients were free of disease at the time of the latest
followup and were without any substantial long-term sequelae. Five patients
had recurrent/metastatic disease, and the mean time to recurrence for these
patients was 1.4 years (range, 0.3 to two years). The sites of recurrence
included bone marrow in two patients and bone, mediastinum, and testes in one
each. Four of the five patients who had a relapse died, and only the patient
with a relapse in the testes survived. Overall, five patients in the current
study died, with one patient each dying from disease progression with
metastatic disease to the mediastinum, diffuse metastatic disease to liver and
bone with sepsis, diffuse metastatic disease with bilateral pulmonary
involvement, chemotherapy-induced acute myelogenous leukemia and sepsis, and
reasons unrelated to the primary diagnosis. One patient who had development of
recurrent disease was successfully treated and survived, whereas one patient
in whom the disease was cured died of other reasons. The mean time from
diagnosis to death for these five patients was 2.1 years (median, 1.9 years;
range, 0.4 to 4.9 years).
Two of ten male patients and three of five female patients died. Four of
the six patients nine years of age or less died, whereas one of the nine
patients ten years of age or older died; the latter patient died from causes
unrelated to the primary diagnosis. None of the eight patients with a delay in
diagnosis of more than three months died, whereas all four patients who died
of disease progression or a treatment-induced neoplasm had been diagnosed
three months or less after the onset of symptoms.
Two of the nine patients with large-cell histology died, while three of the
six patients with non-large-cell histology died. Three of the eight patients
with solitary-bone involvement and two of the seven patients with
multiple-bone disease died. Three of the four patients with radiographic
evidence of a pathologic fracture died, whereas only three of the eleven
patients without evidence of a pathologic fracture died. Two of three patients
with elevated serum calcium died, whereas three of twelve patients without
elevated calcium died of disease progression. Three of the nine patients who
had been managed with chemotherapy alone died, whereas two of the six patients
who had been managed with combined treatment died. Three of seven patients who
had toxicity necessitating a change in treatment died, whereas two of eight
patients who did not require a change in treatment secondary to treatment
toxicity died.
Univariate Analysis
The association of various factors with the risk of death and relapse is
presented in Tables I and
II. The only factor with a
negative prognostic value for survival in the analysis of the fifteen patients
was an age of nine years or less (p < 0.05). Non-large-cell histology,
positive bone-marrow examination, elevated serum calcium, an age of six years
or less, an age of twelve years or less, and female gender were factors that
had a trend toward decreased survival but were not significant risk factors.
The type of treatment, the type of bone (i.e., long bones, flat bones,
vertebral bones), multiple-bone involvement, radiographic evidence of a
soft-tissue mass or pathologic fracture, and physical examination findings did
not have any prognostic value in terms of survival.
Summary Analysis
The characteristics of patients from all existing similar case series of
primary lymphoma of bone in children from the present study (fifteen patients)
and the literature (107 patients) are reported in a table in the Appendix
along with selected survivorship curves for specific factors as discussed
below. Of these 122 patients, ninety-one had data that could be used for
further analysis. These data were collected in order to compare our series
with previously reported series. Of the ninety-one patients, forty-five were
managed with chemotherapy alone and forty-six were managed with a combination
of chemotherapy and radiation. Forty-four patients had solitary-bone disease,
and forty-seven had multiple-bone disease. The bones involved in these
patients and in thirty-one additional patients from a study without specific
treatment or follow-up
data5 are summarized
in Figure 3. According to a
modified Murphy staging system, thirty-six patients were classified as having
stage-I (IE) disease and nine were classified as having stage-II disease
(yielding a total of forty-five patients who were classified as having
localized disease), thirty-two patients were classified as having stage-III
(diffuse) disease, and fourteen patients were classified as having stage-IV
(disseminated) disease.
Twenty-seven of the ninety-one patients died. In this group, age
(regardless of the cutoff) was a predictor of survival. Patients who were
twelve years of age or less, nine years of age or less, and six years of age
or less all had poor survival when compared with older patients (p < 0.05
for all three analyses). In addition, advanced stage, multiple-bone
involvement, and non-large-cell histology were predictive of poor survival (p
< 0.05). The type of bone involved, gender, and the type of treatment did
not have prognostic value for survival.
Eighty-four of the ninety-one patients had a defined histologic subtype.
Seventy-two of these eighty-four patients had either a large-cell or
lymphoblastic subtype, and the remaining twelve had another histologic
subtype. Among patients who were nine years of age or less, twenty-one of
twenty-four patients had a lymphoblastic subtype whereas the other three had a
large-cell subtype. In contrast, among patients who were ten years of age or
older, thirty-four of forty-eight patients had a large-cell subtype and the
other fourteen had the lymphoblastic subtype. The ratio of solitary to
multiple-bone disease was similar across age-groups. Among patients who were
nine years of age or less, seventeen patients had solitary disease and
seventeen had multiple-bone disease. Among patients who were ten years of age
or older, twenty-seven patients had solitary disease and thirty patients had
multiple-bone disease. Among patients who were nine years of age or less,
fifteen of thirty-four patients had localized (Murphy stage-I and II) disease
whereas nineteen had disseminated or diffuse (Murphy stage-III and IV)
disease. Among patients who were ten years of age and older, thirty of
fifty-seven patients had localized disease whereas twenty-seven had
disseminated or diffuse disease.
Clinical Presentation
Primary lymphoma of bone in children is rare compared with other bone
tumors. In a study of seven pediatric cases of primary lymphoma of bone that
were treated over a twenty-five-year period, Howat et al. reported that
eighty-eight cases of Ewing sarcoma were found over the same
time-period4. In the
present study, primary lymphoma of bone was diagnosed in fifteen (4.9%) of 306
children who presented to this institution with non-Hodgkin's lymphoma over a
thirty-three-year period. This finding is similar to other reports, in which
2% to 9% of cases of non-Hodgkin's lymphoma in children involve a bone as the
primary site (see
Appendix)1-3,5,7.
Primary lymphoma of bone affects children of all ages but most commonly is
diagnosed in early adolescence. This was true in our study as well as in other
similar
series1-5,7,17.
Primary lymphoma of bone in children has a predominance in
males1-7.
Younger age was associated with an increased risk of death in our study and
in the summary analysis. Any conclusion from such an analysis has limits, and
it could be argued that the increased mortality in younger children is
secondary to confounding factors. Although multiple-bone involvement was a
predictor of poor survival, solitary disease was as prevalent as multiple-bone
disease in this age-group. However, younger patients were more likely to
present with non-large-cell histology and with disseminated disease, and both
were found to be predictors of poor survival.
The femur, pelvis, and tibia are the bones that are most commonly affected
by this
disease1-7.
However, the type of bone involvement had no prognostic value in terms of
survival or relapse. It appears that patients with primary lymphoma of bone
are equally divided between those with solitary-bone involvement and those
with multiple-bone disease. Although it did not appear that patients with
multiple-bone involvement had a poor prognosis in the current study,
multiple-bone involvement was a significant predictor of poor survival in the
summary study.
The orthopaedic surgeon is likely to be the first to see a child who has a
primary lymphoma of bone because the clinical presentation often includes
localized bone
pain1,7
and occasionally includes a palpable
mass1,2,7.
However, the clinical presentation is often nonspecific, and the clinical
history may not be helpful for making an accurate
diagnosis6. While
most patients in our series complained of pain and the physical examination
often revealed localized swelling or tenderness, the examination revealed
completely normal findings in three patients and there were no physical
findings that had significant prognostic value. Occasionally, patients with
primary lymphoma of bone may present with systemic symptoms such as weight
loss, fever, anorexia, or malaise, which may indicate systemic
spread1,7.
Primary lymphoma of bone is a difficult diagnosis to make without a high
level of suspicion. Therefore, the initial clinical impression is often
different from the final diagnosis and there may be an extended period before
the correct diagnosis is
made1,6,7.
In a study of eleven patients by Furman et al., non-Hodgkin's lymphoma was
considered in only one
patient7. Patients
may experience an extended period of symptoms before presenting to a
physician. In the study by Furman et al., patients sought medical attention
after having had symptoms for a mean of 4.5
months7. Difficulty
related to the histologic evaluation commonly delays an accurate diagnosis,
and a second procedure may be needed if the first is nondiagnostic. This was
true for three patients in our study, and, in a series by Coppes et al., nine
of fourteen patients required a second
procedure1. Even if
adequate pathologic material is obtained, the histologic diagnosis may be
difficult to make, as exemplified by the inaccurate initial diagnosis made for
five of our patients. Molecular methods may be used to confirm the diagnosis;
in one of our patients, the diagnosis was based on molecular criteria alone.
Another complicating factor in making an accurate diagnosis is that there is a
gray and shifting line that separates non-Hodgkin's lymphoma with marrow
involvement from acute lymphoblastic leukemia with a bone lesion. The reason
for the delay varies considerably among patients and is dependent on both
patient and physician attitudes and actions. Separating these factors is
nearly impossible and, although a delay in diagnosis is common, this delay
does not appear to result in poor outcomes.
Laboratory Abnormalities
It has been reported that hypercalcemia is often associated with primary
lymphoma of bone and that it may indicate a poor prognosis. In the study by
Coppes et al., six of fourteen patients had hypercalcemia and five of them
died1. Although the
mortality rate in the present study was higher among patients with elevated
serum calcium than among patients with normal serum calcium, only three
patients had hypercalcemia and this relationship was not found to be
significant. Other laboratory abnormalities have not helped in determining the
diagnosis or prognosis for children with primary lymphoma of bone.
Staging
Studies that have considered primary lymphoma of bone in children have
differed with regard to the types of patients that they have included, and
staging is ambiguous in patients with multifocal bone disease without
bone-marrow
involvement1-5,7.
Some studies have included patients with only localized
disease5, others
have included patients with multiple-bone lesions but have excluded those with
bone-marrow
involvement2,3,
and others have included patients with a primary bone lesion with bone-marrow
involvement1,4,7.
The inclusion criteria used in those studies are important because the
reported prognosis depends on whether patients with disseminated disease are
included, as such patients have worse
outcomes1,4,7.
With the technological advance of staging methods, it now seems acceptable
to consider those with a primary bone lesion with bone-marrow involvement
without other evidence of systemic spread to have disseminated primary
lymphoma of bone1.
It becomes less clear, however, for patients who were staged in the past with
use of inferior imaging methods. It is also difficult to determine the
presence of disseminated disease when the bone-marrow biopsy site is at the
site of disease. In four of the five patients with bone-marrow involvement in
the present series, the ilium was a site of disease, and it is unclear whether
the marrow involvement at this site truly represents disseminated disease.
Imaging
While several studies have concentrated on treatment and outcome, few
reports have discussed the imaging of primary lymphoma of bone in
children1,6,7.
Some investigators have claimed that imaging is most useful for documenting
the extent of
disease6,7.
Radiographic findings are generally nonspecific and can mimic the appearance
of an aggressive lesion such as Ewing sarcoma or may appear to indicate a
benign condition such as osteomyelitis or eosinophilic
granuloma1.
Plain radiographs often demonstrate osteolysis or
osteosclerosis1,2,7
but may fail to demonstrate a lesion at all. On computed tomography, the
majority of lesions appear radiolucent, permeative, or destructive, with or
without sclerosis. Computed tomography studies are more sensitive than plain
radiographs are; in the present study, all eight computed tomography studies
were positive, including those for both patients for whom plain radiographs
were negative. Also, in two patients, computed tomography demonstrated a
soft-tissue mass that was not detected on plain radiographs. Magnetic
resonance imaging of the primary lesion consistently demonstrated low signal
intensity on T1-weighted images and a hyperintense appearance on T2-weighted
images and was more sensitive than computed tomography for detecting a
soft-tissue mass, indicating that magnetic resonance imaging should be
routinely used in the orthopaedic workup of these lesions.
While it could be argued that radiographic characteristics such as a
soft-tissue mass or pathologic fracture might indicate a more aggressive
lesion, neither of these findings was a significant predictor of a poor
prognosis in the present study.
Histologic Analysis
Histologic analysis is difficult, and the differential diagnosis of a
malignant small round-cell infiltrate of bone in a child should include Ewing
sarcoma, rhabdomyosarcoma, lymphoma, neuroblastoma, and primitive
neuroectodermal tumor. Although Langerhans-cell histiocytosis and
osteomyelitis may present similarly, cell-marker analysis will differentiate
these entities7.
Diffuse large-cell or histiocytic lymphoma is the most common histologic
subtype of primary lymphoma of bone, and lymphoblastic lymphoma is also
frequently
encountered1-3,5-7.
Proper histologic analysis is important as some studies have suggested that
histologically driven treatment may result in better
outcomes2, and
non-large-cell histology indicated a poor prognosis for survival in our
summary analysis.
Studies of primary lymphoma of bone in children have included limited
information on
immunophenotype1,5,
although it has been suggested that this factor may have clinical
importance2.
Hutchison et al. demonstrated that immunological patterns are prognostic risk
factors in children with non-Hodgkin lymphoma and that B-cell phenotype may be
associated with a better
prognosis18.
Although the current series offers the most extensive and complete
immunophenotype analysis of primary lymphoma of bone in children to date, the
prognostic significance of immunophenotype could not be assessed in the
present study because all patients with large-cell lymphoma demonstrated a B
cell type and all patients with lymphoblastic lymphoma demonstrated a
precursor B subtype (Fig.
4).
Treatment/Outcomes
Non-Hodgkin's lymphoma in children is a systemic disease; therefore, local
therapy is not sufficient, even if the condition presents as localized
disease6. Treatment
with combination chemotherapy regimens has improved the prognosis associated
with non-Hodgkin's lymphoma in this
age-group19-22,
and the efficacy of chemotherapy for the treatment of primary lymphoma of bone
in children is well
established23.
Serious therapy-related late effects are a problem in the treatment of
children7. Radiation
therapy may lead to compromised musculoskeletal development and the appearance
of secondary malignant
disease5. This risk
is exemplified by the reported appearance of secondary bone tumors in two
patients, five and 7.5 years after the use of radiation therapy for the
treatment of primary lymphoma of
bone3. Both tumors
occurred in the radiation field at the site of the original treatment. Similar
findings were not reported in a study of thirty-one patients who were followed
for six to twelve years after the remission of
disease5 or in
another series of thirty-one patients with long-term
follow-up2. In our
experience, one patient had development of a chemotherapy-induced acute
myelogenous leukemia.
Several series have demonstrated excellent survival in patients who have
been managed either with chemotherapy alone or with chemotherapy combined with
radiation therapy, suggesting that chemotherapy alone is likely adequate for
this
disease2,3.
In our series and in the summary analysis, patients who had been managed with
chemotherapy alone had similar outcomes compared with those who had been
managed with concurrent radiation therapy. Given the risk of growth
disturbances, organ dysfunction, and secondary malignant disease after
radiation therapy, it seems logical to avoid its use. It is important to
realize that patients in the present study and in similar studies were managed
differently over time, which makes interpretation of the data difficult.
The mean duration of follow-up for the ten surviving patients in the
present study was 13.6 years (range, two to twenty-three years). This
represents a longer follow-up period than has been reported for any previous
series. As increasing numbers of survivors of childhood cancer live into
adulthood, the prevalence of long-term complications is expected to increase,
and it is therefore important to document long-term
follow-up5.
The mortality rate that is reported depends on what types of patients are
included1,4,7.
Studies that include patients with only localized disease demonstrate a better
survival5. In a
study of thirty-one patients, the overall product-limit-estimated five-year
event-free survival rate was 84%, with the event-free survival rate for
seventeen patients with localized disease being 94% and the event-free
survival rate for patients with more advanced disease being
70.7%2. Overall,
studies have demonstrated a survival rate of between 40% and 100% for children
with primary lymphoma of
bone1-7,with
an event-free survival rate of 75% to 100% for those with localized disease
and 25% to 71% for those with disseminated
disease2. In our
summary analysis, patients with advanced-stage disease had a lower survival
rate. The survival of patients in the present study is consistent with the
relevant literature because although five of the fifteen patients died, the
present study included patients with localized, diffuse, and disseminated
disease.
Limitations
Studying primary lymphoma of bone in children is challenging because it is
a rare disease and, in order to obtain a large case series, it is necessary to
collect cases over several decades. In addition, the approaches to diagnosis,
staging, and treatment vary over time. Advances in chemotherapy have
substantially improved outcome in recent years, and including patients who
were managed thirty years ago along with patients who were managed in the past
five years may cloud conclusions about prognosis. Because of this variation,
it could be postulated that patients who are managed with more modern
therapeutic regimens are likely to have a better outcome; however, this is
difficult to prove in a study of this size.
It should be noted that our univariate analysis of patients in this series
had two limitations. First, the number of patients from our institution was
small and may have failed to detect key trends. Second, although we tried to
identify potential confounding factors, it was not possible to evaluate
confounding factors in a univariate analysis involving a small number of
patients. There are also inherent problems associated with combining data from
the literature, and any conclusions about prognosis that are made on the basis
of data obtained in this way are informational but certainly must be
considered with great care.
It could be argued that the results of the present study are not much
different from the natural history of this disease in adults. Although this
may be true, pediatric patients have unique risks associated with treatment
and disease that are not present for adults, and describing this population
individually is important. In order for clinicians to better understand this
entity, large series of patients with this condition who have been treated
with current therapy and have had long-term follow-up are necessary.