By continuing to browse this site you agree to us using cookies as described in About Cookies
Notice: Wiley Online Library will be unavailable on Saturday 7th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 08.00 EDT / 13.00 BST / 17:30 IST / 20.00 SGT and Sunday 8th Oct from 03.00 EDT / 08:00 BST / 12:30 IST / 15.00 SGT to 06.00 EDT / 11.00 BST / 15:30 IST / 18.00 SGT for essential maintenance. Apologies for the inconvenience.
The low incidence of primary lymphoma of bone (PLB) has led to discrepancies in classification as well as difficulty in prognostication. The authors of this report used the Surveillance, Epidemiology, and End Results (SEER) database to analyze a large, population-based cohort of adult patients with this disease. The database provides a standardized classification and documentation of outcomes and enables a meaningful evaluation of prognostic factors.
The SEER database was used to identify all patients who were diagnosed with PLB from 1973 through 2005. Survival was analyzed with the Kaplan-Meier method, and the influence of clinical parameters on survival was analyzed with the log-rank test. A Cox proportional hazards model was used for multivariate analysis.
Fifteen hundred adult patients with PLB were analyzed. The 5-year and 10-year survival rates for adult patients were 58% and 45%, respectively. Multivariate analysis revealed that younger age and localized disease were independent predictors of survival. It is noteworthy that the incidence of disease, as determined by the annual percentage change, increased during the study period (P < .05).
Primary lymphoma of bone (PLB) is a rare disease that was described first by Oberling in 1928.1 However, it was only after Parker and Jackson reported their series that it was recognized as a distinct clinical entity.2 PLB has been described as a malignant, lymphoid infiltrate within bone, with or without cortical invasion or soft tissue extension, and without concurrent involvement of regional lymph nodes or distant viscera.3, 4 The definition of PLB is controversial throughout the literature. Some reports have included only patients with Ann Arbor stage I and stage II disease,5, 6 whereas others have included patients with stage IV disease (involvement of bone marrow).7-12 In addition, some authors have included patients with lymph node involvement in the diagnosis of PLB.13-15 In contrast to adults, PLB in children is regarded as a clinical entity that is distinct from its adult counterpart. Like many other pediatric malignancies, PLB is considered a systemic disease in children.16-18 In small retrospective studies, PLB in children is characterized by rapid progression, a higher incidence of micrometastasis, and a propensity for spread to the central nervous system, yet children have a better prognosis.19, 20
Non-Hodgkin lymphoma (NHL) constitutes the majority of PLBs, and the most common subtype is diffuse large B-cell lymphoma.8, 11 PLB constitutes only 2% of all bone tumors and 5% of all extranodal lymphomas.21, 22 With improvement in staging procedures, especially with the combined use of computed tomography (CT), magnetic resonance imaging (MRI), and now positron emission tomography (PET), the proportion of patients diagnosed with systemic disease (stage IV, Ann Arbor classification) has increased. Traditionally, most patients present with limited stage disease (stages I and II) in the fifth or sixth decade of life.23 Because of the infrequency of the disease, prospective, randomized clinical trials addressing treatment options have been few. Radiation therapy has been established as a standard of care for the local control of PLB since the 1960s. Radiation therapy results in adequate local control, but patients subsequently exhibit relatively high rates of systemic disease recurrence. Recently, there has been some evidence of improved survival with combined systemic therapy using rituximab and combination chemotherapy with cyclophosphamide, doxorubicin, vincristine, and prednisone.8 Most reports of improved survival have been individual center-based and have included a limited sample size. The limited number of patients in these studies restricted their value in identifying the prognostic factors for survival. The rarity of the disease, combined with discrepancies in histologic criteria and staging, makes it difficult to compare outcomes between studies. In that regard, the survival rate has differed greatly among individual series, and the reported 5-year survival rate has ranged from 88% to 58% 5-year.7, 9 In an analysis of the Surveillance, Epidemiology, and End Results (SEER) database, we attempted to elucidate survival rates for patients who had a diagnosis of PLB in a population-based database and to identify prognostically significant parameters. By stratifying patients on the basis of extent of disease, we attempted to clarify the definition of what constitutes PLB and what underlies the discrepancies in outcomes reported in the literature. We did not include pediatric patients in our study cohort in an attempt to avoid heterogeneity of the disease process.
MATERIALS AND METHODS
The population that we studied was extracted from population-based cancer registries that participate in the SEER Program, which was established by the National Cancer Institute (NCI) as a direct result of the National Cancer Act of 1971. Currently, SEER collects data from 17 population-based registries that cover approximately 26% of the US population. It is the only comprehensive source of population-based data in the United States that includes information on the stage of cancer at the time of diagnosis with follow-up on all patients for survival data. In addition, each registry collects data on patient demographics, primary tumor site and morphology, and first course of treatment (received within 4 months of diagnosis). The specific local registries were selected for their completeness and adequate representation of minority populations. The SEER Program currently is regarded as the standard of quality among cancer registries around the world and has case completeness of 98%.24
In total, 1559 patients with lymphoma who had bone as their primary disease site were identified from 1973 through 2005 in the SEER database. Patients with lymphoma were identified according to International Classification of Disease for Oncology, 3rd edition (ICD-O-3) histologic codes 9590/3, 9591/3, 9670/3, 9671/3, 9675/3, 9680/3, and 9684/3. Thirteen of these patients were excluded, because information regarding histologic diagnosis was not available. An additional 43 of the remaining patients were in the group ages 0 to 16 years, and they were excluded from further analysis because of the potentially different natural history of disease in that age group. Three patients lacked information in some key areas, such as age, sex, or disease stage, and were censored, leaving a total population of 1500 adult patients for further analysis. Information regarding patient demographics, stage at diagnosis, primary site and size, year of diagnosis, surgical and radiation treatment, and survival until death or loss to follow-up was collected. Percentages were based on available data for each individual variable. Patients who had missing data were excluded from the respective univariate and multivariate analyses.
Patient age was converted arbitrarily into a categorical variable (ages 17-29 years, 30-59 years, and ≥60 years) for the purpose of analysis. Similarly, anatomic tumor location was changed to a categorical variable: Appendicular and axial were used as locations for tumors that were not described uniformly. The appendicular skeleton included long and short bones of limbs and associated joints. The scapula was included in the appendicular skeleton, whereas the pelvic bones and associated joints were categorized under axial skeleton. The axial skeleton included vertebra, ribs, sternum, clavicle and associated joints, bones of the skull and face, associated joints, and the mandible. Staging categories of local, regional, and distant disease were used according to the American Joint Committee on Cancer classification. Year of diagnosis was categorized into 4 categorical variables: 1973 through 1975, 1976 through 1985, 1986 through 1995, and 1996 through 2005. The results reported herein are in compliance with the Health Insurance Portability and Accountability Act of 1996.
The software program SEER*Stat (version 6.4.4; NCI, Bethesda, Md) was used to analyze incidence rates and trends from 1973 through 2005. Incidence rates were age-adjusted and normalized using the 2000 US standard population. The annual percentage change (APC) was calculated using the weighted least-squares method as described in the SEER Cancer Statistics Review (1973-2005). Statistical analyses were performed using the SPSS Statistical package (version 16.0; SPSS Inc., Chicago, Ill). Chi-square tests were used to assess correlations between categorical variables. Overall and disease-specific survival from the time of initial diagnosis to the date of last contact (or the date of death, if the patient died) was calculated using the Kaplan-Meier method. The effects of demographic, clinical, pathologic, and treatment variables were tested using the log-rank test for categorical values. A multivariate analysis was performed to determine independent prognostic factors using a Cox proportional hazards model. All prognostic factors that were identified as significant in the univariate analysis were included in the multivariate analysis. A P value <.05 was considered statistically significant.
Overall, 1500 adult patients with PLB were identified in the SEER database and were included in the current analysis: The demographic characteristics of these patients are displayed in Table 1. The majority of patients in this population were Caucasian and non-Hispanic. Men constituted a slightly greater proportion (53.9%) of the cohort. Most tumors were of the “NHL-large B-cell diffuse” type (66.3%), and most patients underwent surgery and/or received radiation therapy as initial treatment (993 patients vs 503 patients, respectively).
Table 1. Demographic and Clinical Characteristics of the Adult Cohort
Valid % of Total
NOS indicates not otherwise specified; NHL, non-Hodgkin lymphoma.
Total no. of patients
Malignant lymphoma, NOS
Malignant lymphoma, non-Hodgkin, NOS
NHL, small B-cell lymphocytic, NOS
NHL, mixed small and large cell diffuse
NHL, large B-cell diffuse
NHL, large B-cell diffuse immunoblastic
Primary tumor site
No. of lesions
Year of diagnosis
The overall survival of adult patients with PLB is illustrated in Figure 1. The 5-year and 10-year survival of these was 58% and 45%, respectively. The incidence of PLB determined according to the APC increased during the study period (3.7%; P < .05) (Fig. 2).
Univariate and multivariate analyses of the parameters that influenced survival are provided in Tables 2 and 3, respectively. Univariate analysis demonstrated a survival advantage for adult patients who were younger (P < .001), had local stage disease with no systemic disease involvement (P < .001), had appendicular disease locations (P < .001), had higher grade tumors (P = .02), had “NHL-large B-cell diffuse” histology (P = .011), and received radiation therapy for local disease control (P = .04). Because the majority of patients in the cohort had high-grade histology (97%), the association between improved survival and grade may be spurious, because the other grades had low representation. Patients who received treatment between 1973 and 1975 had significantly poorer outcomes compared with patients who received treatment in the later decades, although the number of patients (n = 27) was limited. There was no significant difference in survival among patients who received treatment over the subsequent 3 decades (Table 2).
Table 2. Disease-Specific Survival According to Demographic and Clinical Characteristics (Proportion Surviving): Univariate Survival Analysis for Adult Patients With Primary Lymphoma of Bone
NA indicates not available; NOS, not otherwise specified; NHL, non-Hodgkin lymphoma.
P values are shown for the log-rank test between variables.
P = .178 only for white versus others; P = .249 for white versus black; and P = .858 for black versus others.
P = .011 for NHL, large B-cell diffuse versus NHL, small B-cell lymphocytic NOS only; P = .920 for malignant lymphoma NOS versus malignant lymphoma, non-Hodgkin, NOS; P = .290 for malignant lymphoma NOS versus NHL, small B-cell lymphocytic NOS; P = .665 for malignant lymphoma NOS versus NHL lymphoplasmacytic; P = .374 for malignant lymphoma NOS versus NHL mixed small and large cell diffuse; P = .037 for malignant lymphoma NOS versus NHL, large B-cell diffuse; P = .690 for malignant lymphoma NOS versus NHL, large B-cell diffuse immunoblastic; P = .517 for malignant lymphoma, non-Hodgkin, NOS versus NHL small B-cell lymphocytic; P = .807 for malignant lymphoma, non-Hodgkin, NOS versus NHL lymphoplasmacytic; P = .372 for malignant lymphoma, non-Hodgkin, NOS versus NHL mixed small and large cell diffuse; P = .022 for malignant lymphoma, non-Hodgkin, NOS versus. NHL large B-cell diffuse; P = .702 for malignant lymphoma, non-Hodgkin, NOS versus NHL large B-cell diffuse immunoblastic; P = .642 for NHL, small B-cell lymphocytic NOS versus NHL lymphoplasmacytic; P = .123 for NHL, small B-cell lymphocytic NOS versus NHL mixed small and large cell diffuse; P = .603 for NHL, small B-cell lymphocytic NOS versus NHL large B-cell diffuse immunoblastic; P = .385 for NHL lymphoplasmacytic versus NHL mixed small and large cell diffuse; P = .150 for NHL lymphoplasmacytic versus NHL large B-cell diffuse; P = .942 for NHL lymphoplasmacytic versus NHL large B-cell diffuse immunoblastic; P = .783 for NHL mixed small and large cell diffuse versus NHL large B-cell diffuse; P = .427 for NHL mixed small and large cell diffuse versus NHL large B-cell diffuse immunoblastic; and P = .041 for NHL large B-cell diffuse immunoblastic versus NHL large B-cell diffuse.
P < .001 for localized versus diffuse only, P = .611 for localized versus regional, and P = .044 for regional versus diffuse: P < .044.
P < .001 for 1973 through 1975 versus 1986 through 1995 and 1996 through 2005 only, P = .003 for 1973 through 1975 versus 1976 through 1985, P = .812 for 1976 through 1985 versus 1986 through 1995, P = .673 for 1976 through 1985 versus 1996 through 2005; and P = .086 for 1986 through 1995 versus 1996 through 2005.
In multivariate analysis, adult patients who had local disease and who were younger had a significant survival advantage. Thus, age (Fig. 3) and extent of disease (Fig. 4) constituted independent predictors of survival among adult patients with PLB. Further analysis revealed that there was an association between younger age and appendicular disease location: In the group ages 17 to 29 years, there were 120 appendicular disease sites and 47 axial disease sites; whereas, in the group aged ≥60 years, there were 293 appendicular disease sites and 548 axial disease sites (P < .001). Similarly, there was a significant association between younger age and a histologic classification of “NHL-large B-cell diffuse,” which explained the loss of significance for histologic classification in subsequent multivariate analysis (data not shown; P = .002). Local disease stage also was associated significantly with the receipt of radiation therapy (P < .001; data not shown). A survival analysis of patients with localized disease revealed that radiation therapy was associated with improved survival (P = .038) (Fig. 5).
The current study presents an analysis of a population-based cohort of patients with a diagnosis of PLB compiled from the treatment at 17 registries across the United States. Previous reports presented retrospective case series with limited numbers of patients and nonuniform treatments and outcome measures. The population-based nature of the SEER database is important, because this normalizes the outcome analysis from potential selection biases and standardizes both classification and outcome criteria. An additional strength of the SEER database is the parameter case completeness, which is 98%; thus, variability in reporting is not a factor.24
The diagnosis of what constitutes PLB has been controversial. The current study focused on a diagnosis of lymphoma and a primary location in bone as defined by the ICD-O-3. The histologic classification that we used in the current study was determined by what is included in the SEER database. The ICD-O-3 codes for the histologic subtypes that we included in the current analysis are indicated above (see Materials and Methods). However, the evolution of classification and staging for this disease over the course of the period of study (1973-2005) is a limitation of the database structure; if this database is to be queried for PLB, then this limitation must be accepted. Further limitations of the SEER database include the lack of specifics regarding chemotherapeutic treatments and the nature of the therapeutic information, which is limited to only the first treatment after diagnosis. Another limitation of this study is that the SEER database did not allow us to analyze other objective prognostic factors (such as lactate dehydrogenase levels), other laboratory parameters, or the International Prognostic Index score as variables in survival for patients with PLB. With these as acknowledged limitations, the SEER database is internally consistent, and all patients who are included are classified with the same criteria. The controversy in classification is probably 1 of the contributory factors for the discrepancy in outcomes that have been reported for patients with PLB.
It is noteworthy that the decade in which treatment was rendered did not seem to influence outcome in the current study. This contradicts the results of other studies, in which it was reported that outcomes improved with newer, multimodal chemotherapeutic regimens, including rituximab.8 The database was limited for this interpretation, because only the initial therapeutic regimens are reported in SEER; therefore, it is not certain whether patients adhered to a specific protocol over time or received subsequent treatments later in the disease process. In addition, patients who were included in the SEER database may have discontinued the initial protocol or may have received outdated therapies. We realize that centers with standardized protocols report better survival outcomes than those reported in this population-based cohort of multiple centers.7, 8 These patients presumably all were treated on contemporary protocols at the time of their diagnosis in tertiary treatment centers, suggesting that the initial treatment was somewhat consistent. These findings stress the importance of defining an optimal treatment strategy for this disease entity to improve the outcome for all patients who are diagnosed with PLB independent of the location of their treating institution.
The 5-year and 10-year survival rates of adult patients in the current study were 58% and 47%, respectively. These rates are comparable to earlier results reported from smaller series, which often were from a single institution or region (Table 4), except for the series that originated from the Memorial Sloan-Kettering Cancer Center, in which the reported 5-year survival rate was 88%. That analysis also indicated an improved survival among younger adult patients, similar to previous studies.7, 8, 12 In the current study, we did not evaluate PLB in the pediatric age group because of the small number of patients 43; therefore, we cannot comment on the better survival in pediatric patients versus adult patients with lymphoma that has been suggested by others.15
Table 4. Comparison of Previous Studies
No. of Patients
Type of Survival
5-Year Survival, %
SEER indicates Surveillance, Epidemiology, and End Results Program.
Patients with locally confined disease showed a significantly better survival prognosis than those with more extensive disease. This subgroup of patients might be considered a separate clinical entity from adult patients with systemic staging of PLB. Patients with only localized disease may constitute a distinct clinical population that might be amenable solely to local treatment strategies such as radiation and/or surgery. Surgery did not prove to be of survival benefit in this study although the specifics of surgeries and surgical strategies are not specified in detail in the SEER database. It is noteworthy that radiation therapy was associated with improved survival of patients with locally limited disease (Fig. 5). Local disease and the utilization of radiotherapy were correlated which may explain the loss of survival advantage associated with radiation therapy on multivariate analysis. Only a study addressing therapeutic options in the specific cohort of patients with locally limited PLB could further clarify if surgery and/or radiation therapy in and of themselves might alter the prognosis.
The APC in the incidence of PLB increased during the study period, and the reason behind this is unclear. A possible explanation could be an increase in the recognition of this disease entity over time. Alternatively, environmental and viral exposures that reportedly influence the rise of other lymphoma types also may play a role in the increased incidence of PLB over time.7, 25, 26
Recently, Hodgkin lymphoma histology has been included in the classification of PLB, although the corresponding group of patients with this histologic subtype numbered only 6 in the SEER database. Thus, we did not include patients with Hodgkin disease in the current study for analysis. The histologic subtype “NHL-large B-cell diffuse” was associated significantly with improved survival in univariate analysis along with younger age. The survival advantage associated with histologic subtype lost its significance on multivariate analysis, which can be attributed to the strong correlation between this histology and younger age. Similarly, the loss of a survival advantage associated with an “appendicular” location of disease in multivariate analysis may be attributed to the strong correlation between that disease location and younger age.
PLB was defined as a separate disease entity in the realm of lymphoma on the notion of a comparably favorable outcome. Classification of this disease entity became controversial over time because of improved diagnostic modalities. With the emergence of CT, MRI, and PET, patients who were diagnosed with solitary lymphoma lesions in bone often demonstrated disease in other regions, which made the label PLB more and more controversial. The current study demonstrated that patients with localized PLB had a better outcome than patients with other types of NHL. The 5-year and 10-year overall survival rates for patients with NHL, according to the SEER database, were 53% and 43%, respectively.27 Comparing these values with the 5-year and 10-year survival rates for all patients with PLB revealed little difference, if any, among patients who had systemic disease (Table 2). Conversely, patients with localized disease had much better 5-year and 10-year survival rates (65% and 53%, respectively) in the current study. This subset of patients benefited from external-beam radiation as a local treatment measure. The current study supports subclassifying this group of patients with PLB who have localized disease, because there is a difference in their prognosis and outcome compared with patients who have other, more extensive disease. All other patients with bone lymphoma likely have disease that constitutes a systemic process with a worse prognosis.
Historically, there have been controversies about to the classification of PLB; and, to our knowledge, this study is the first to address a large, population-based cohort of adult patients with PLB. Our results emphasize the need to differentiate between locally limited PLB and the potentially systemic variant. These 2 populations have different characteristics with regard to clinical outcomes. The current study suggests that treatment for localized PLB may differ appropriately from treatment for systemic lymphoma. Local disease can be treated effectively and successfully with radiation therapy, whereas the role of surgery in local disease needs further clarification. On the basis of the current study, future studies should analyze local PLB and systemic PLB in adults separately, and we recommend limiting use of the name “PLB” only to truly local disease with a single osseous lesion. The overall prognosis for patients with systemic PLB is poor, and treatment modalities will have to be improved and standardized across centers to improve the prognosis for these patients regardless of treatment location. The current study documents survival in a large cohort of patients that can be used as a historic control for future therapeutic trials.