Using primary site as a predictor of survival in mantle cell lymphoma


  • Alexander J. Ambinder,

    1. Department of Hematology and Medical Oncology, Winship Cancer Institute, Emory University, Atlanta, Georgia
    Search for more papers by this author
  • Pareen J. Shenoy MBBS, MPH,

    1. Department of Hematology and Medical Oncology, Winship Cancer Institute, Emory University, Atlanta, Georgia
    Search for more papers by this author
  • Loretta J. Nastoupil MD,

    1. Department of Hematology and Medical Oncology, Winship Cancer Institute, Emory University, Atlanta, Georgia
    Search for more papers by this author
  • Christopher R. Flowers MD, MS

    Corresponding author
    1. Department of Hematology and Medical Oncology, Winship Cancer Institute, Emory University, Atlanta, Georgia
    • Emory University, Winship Cancer Institute, 1365 Clifton Road NE, Building B, Suite 4302, Atlanta, GA 30322

    Search for more papers by this author
    • Fax: (404) 778-3366



Mantle cell lymphoma (MCL) is a rare B cell lymphoma that varies in clinical behavior with some patients experiencing aggressive disease with short survival, whereas others have indolent behavior. We examined the association between primary disease site and survival in MCL patients to identify subgroups with distinct characteristics.


We analyzed the United States Surveillance, Epidemiology and End Results Program database for MCL cases reported from 2000 through 2009. Kaplan-Meier curves and Cox proportional hazard models were used to estimate the effect of primary site on survival.


Among 4477 cases included in our study, 19.6% of patients presented with an extranodal primary site. The most common extranodal primary sites were of the gastrointestinal (GI) tract (7.8%), the head and neck (6.2%), and the hematologic/reticuloendothelial systems (3.6%). Asians/Pacific Islanders were more likely than whites or blacks to have GI tract or head and neck disease (P < .0001 and P = .002, respectively). Advanced disease and B symptoms were less common in those with primary disease of the GI tract or head and neck than in those with primary disease of the lymph nodes (both P < .0001). In a multivariate Cox regression model, patients with primary disease of the GI tract and head and neck had superior survival compared to those with primary disease of the lymph nodes; hazard ratios 0.75 (95% CI = 0.62-0.90) and 0.68 (95% CI = 0.55-0.85), respectively.


Primary site of disease may be an important prognostic factor for patients with MCL. Further studies elucidating a biological basis for these differences are needed. Cancer 2013. © 2013 American Cancer Society.

Mantle cell lymphoma (MCL) is an uncommon but distinctive and aggressive subtype of non-Hodgkin lymphoma (NHL). MCL was formally recognized as a distinct subtype of NHL in 1992 and was incorporated into the Revised European-American Classification of Lymphoid Neoplasms (REAL) in 1994 and later by the World Health Organization (WHO).1, 2 With improvements in the classification systems that included the addition of data on morphology, immunophenotype, genotype, stage of differentiation, and clinical features, meaningful epidemiologic studies of NHL subtypes such as MCL can now be performed. Illustrating the importance of these improvements in diagnosing and reporting lymphomatous disease, the International Lymphoma Study Group published a study analyzing the effect of the inclusion of additional diagnostic data on diagnostic accuracy. For MCL, the inclusion of immunophenotypic data improved expert pathologist agreement with the consensus diagnosis from 77% to 87%.3

MCL is more common in men than in women, whites than in blacks, and approximately 75% of cases present with advanced stages of disease.4 The median overall survival (OS) ranges from 3 to 5 years, with poorer survival being associated with advanced age, male sex, and advanced stage of disease.5 Although MCL can be an aggressive disease, a subset of patients have an indolent clinical course with survival lasting more than 10 years.6 The most clinically useful prognostic factors for survival are those that constitute the MCL International Prognostic Index (MIPI) score including: patient age, performance status, lactate dehydrogenase levels, and white blood cell count. The MIPI was developed using data collected by the European MCL Network from 400 patients with MCL7 and has been validated in several studies.8-11 In addition, other candidate prognostic factors have been identified, including the Ki-67 index and p53 mutation status.12-14

Recognizing patient subgroups with longer expected survival would likely influence clinical decision-making, but a definitive marker of indolent MCL has yet to be discovered. Recent studies have suggested that non-nodal disease may be associated with improved survival.15-18 The purpose of this study is to analyze the United States Surveillance, Epidemiology and End Results (SEER) data set to examine the association between primary site of disease and survival outcomes in patients with MCL.


Data Source

The SEER Program collects data on cancer incidence and survival from population-based registries throughout the United States. The database has expanded from 9 registries in 1973 (5 rural and 4 metropolitan) to include 18 registries that sample approximately 28% of the United States population.19, 20 For our analyses, we used the 2000 through 2009 data from the SEER 18 registries.


The SEER classification system for lymphoid neoplasms has undergone several revisions since its inception. From 1973 through 1977, lymphoid neoplasms were classified according to the Manual of Tumor Nomenclature and Coding.21 In 1978, SEER adopted the International Classification of Diseases for Oncology (ICD-O) coding system.22 In 1992, SEER updated its classification of lymphoid neoplasms to the ICD-O-2 system.23 In 2001, the WHO classification, which combines aspects of the REAL classification and the French-American-British classification, was introduced. Recently, SEER adopted the ICD-O-3 coding system and devised a formula for converting ICD-O-2 codes into ICD-O-3 codes.

We identified MCL cases using ICD-O-3 histology code 967324 in accordance with the InterLymph Consortium classification of lymphoid neoplasms for epidemiologic research based on the 2008 WHO classification.25, 26 Exclusion criteria were: patients of unknown age or age < 18 years, a diagnosis of MCL confirmed only by death certificate, patients who were not actively followed by SEER, patients for whom the diagnosis of MCL was a secondary or later primary, and patients with unknown primary site. All data refer to the incidence of neoplasms with malignant behavior. Figure 1 illustrates the selection of the study cohort.

Figure 1.

The figure provides an overview of the study cohort with reasons for inclusion/exclusion through the selection process. The text with dashed boxes on the right denotes the cases excluded. Abbreviations: dx, diagnosis; ICD-O-3, International Classification of Diseases for Oncology, third edition; MCL, mantle cell lymphoma; SEER, Surveillance, Epidemiology and End Results data set.

Data regarding demographics, tumor morphology and stage, the presence of B symptoms, extranodal involvement, primary site, treatment (radiation and surgery), and survival were used for this study. Age was categorized according to the MIPI age categories (< 50, 50-59, 60-69, > 69 years). Patient race was recoded as white, black, Asian/Pacific Islander, and “other,” a category that includes American Indian/Alaska Native and subjects of unspecified or unknown race. Disease stage at diagnosis was categorized into localized disease (Ann Arbor stages I and II) and advanced disease (Ann Arbor stages III and IV). Primary sites were concatenated according to organ or anatomic site into 12 categories: 1) head and neck; 2) gastrointestinal (GI) tract; 3) pulmonary system; 4) thymus, mediastinum, and heart; 5) musculoskeletal system; 6) hematologic and reticuloendothelial system (Heme/RES); 7) integumentary system; 8) nervous system; 9) breast tissue; 10) genitourinary system; 11) endocrine system; and 12) lymphatic system. Categories that accounted for < 10% of all extranodal disease were grouped into “Other” primary site. Survival time was calculated using the date of diagnosis and one of the following: date of death, date last known to be alive, or date of the study cutoff (December 31, 2009).

Statistical Analysis

Comparisons of baseline characteristics across sexes, races, and by primary site were made using analysis of variance (ANOVA) and chi-square tests. Kaplan-Meier survival curves were generated and compared using log-rank tests. Univariate and multivariate Cox proportional hazard models were developed to examine the association between primary site and survival. Covariates considered for inclusion in the adjusted models were age at diagnosis, sex, race, disease stage, presence of B symptoms, year of diagnosis, and treatment modalities. Variables for which > 10% of observations were missing data were not initially included in the multivariate model, but were included in sensitivity analyses. A level of significance (alpha) of 0.05 was considered statistically significant. All statistical analyses were performed using SAS software, version 9.3.


Baseline Characteristics

A total of 5724 cases of MCL were recorded between 2000 and 2009; 37 of these cases were confirmed by death certificate alone, 2 were not actively followed, 1199 were second or later primaries, and 8 had unknown primary sites, and thus were excluded from the analysis. The final study cohort contained 4478 cases (Fig. 1).

The most common primary sites of disease were: lymph nodes (80.4%), GI tract (7.8%), head and neck (6.2%), and Heme/RES (3.6%, Table 1). Of patients presenting with extranodal primary sites, 39.7% presented with tumors of the GI tract, 31.7% in the head and neck, and 18.4% had disease of the Heme/RES. Malignancies of the stomach (13.5%), small intestines (12.0%), and colon (51.3%) comprised the majority of cases occurring in the GI tract, whereas diseases of the oropharynx (36.9%) and eye/adnexa (21.5%) accounted for most cases in the head and neck. Among those presenting with disease of the Heme/RES, 53.1% presented with disease in their bone marrow and 46.3% with splenic disease.

Table 1. Frequencies of Primary Sites and Composition of Extranodal Sites
Primary SiteFrequency (%)Extranodal Cases (%)
Lymphatic system3598(80.4)
Gastrointestinal tract349(7.8)39.7
Head and neck279(6.2)31.7
Pulmonary system26(0.6)3
Integumentary system18(0.4)2
Musculoskeletal system17(0.4)1.9
Breast tissue12(0.3)1.4
Genitourinary system7(0.2)0.8
Nervous system3(0.1)0.3
Endocrine system2(0)0.2

Table 2 shows patient characteristics at presentation stratified by primary site of disease. Males comprised 69.7% of the study cohort, and there was no difference in the proportion of males across primary sites. A majority of MCL patients in SEER were white (91.3%). Extranodal primary sites were significantly more common among Asian/Pacific Islander patients than either whites or blacks (P < .0001 and P = .002, respectively). Patients with primary disease sites of the GI tract (56.9%) and head and neck (40.5%) less commonly presented with advanced disease (Stages III/IV) at diagnosis compared with those with primary disease of the lymph nodes (86.8%, ANOVA P < .0001). Similarly, fewer patients with primary disease of the GI tract (26.2%) and head and neck (12.9%) presented with B symptoms at the time of diagnosis compared to those with primary disease of the lymph nodes (34.3%, ANOVA P < .0001). The proportions of subjects receiving surgical and radiation therapy also varied across different primary sites. Although 28.3% of patients with disease of the head and neck received radiation treatment, only 8.5% among those presenting with primary disease of the lymph nodes received radiation (ANOVA P ≤ .0001), suggesting that the treatment patterns followed the prior findings that head and neck primary sites tended to be localized.

Table 2. Baseline Characteristics and Comparisons Within and Across Primary Site Categoriesa
 OverallLymph NodeGI TractHead and NeckHeme/RESOther 
Variable(N = 4477)(n = 3598)(n = 349)(n = 279)(n = 162)(n = 89)P
  • a

    Numbers in parentheses indicate percentage (%) unless specified.

  • b

    Percentages for other races not shown (n = 22).

  • Abbreviations: GI, gastrointestinal; Heme/RES, hematologic and reticuloendothelial system; SD, standard deviation.

Male3122 (69.7)2507 (69.7)257 (73.6)195 (69.9)106 (65.4)57 (64.0).26
Raceb      <.001
 White4048 (91.3)3258 (80.5)302 (7.5)251 (6.2)152 (3.8)85 (2.1) 
 Black191 (4.3)166 (86.9)13 (6.8)5 (2.6)4 (2.1)3 (1.6) 
 Asian/Pacific Islander171 (3.9)118 (69.0)28 (16.4)19 (11.1)5 (2.9)1 (0.6) 
Mean age at diagnosis, y (SD)66.7 (12.2)66.6 (12.2)67.8 (11.3)66.5 (12.3)67.5 (11.3)68.3 (13.3).23
Age at diagnosis, y      .28
 <50382 (8.5)319 (8.8)22 (6.3)25 (9.0)9 (5.6)7 (7.9) 
 50-59934 (20.9)765 (21.3)60 (17.2)61 (21.9)36 (22.2)12 (13.5) 
 60-691218 (27.2)969 (26.9))111 (31.8)70 (25.1)43 (26.5)25 (28.1) 
 >691943 (43.4)1545 (42.9)156 (44.7)123 (44.1)74 (45.7)45 (50.6) 
Diagnosed before 20052050 (45.8)1666 (46.3)154 (44.1)124 (44.4)67 (41.4)39 (43.8).66
Advanced stage3402 (80.7)2946 (86.8)181 (56.7)107 (40.5)126 (83.4)42 (48.3)<.001
B symptoms present1021 (34.3)898 (36.8)55 (26.2)23 (12.9)36 (37.9)9 (16.7)<.001
Received surgery1625 (36.5)1300 (36.4)115 (33.2)126 (45.5)54 (33.3)30 (33.3).02
Received radiation443 (10.1)307 (8.7)25 (7.3)79 (28.8)5 (3.1)27 (30.7)<.001

Survival Outcomes

The survival of patients with MCL varied by primary site of disease. Patients with primary disease of the lymph nodes had worse survival (median OS = 48 months, 5-year OS = 43%) compared to patients with that of the GI tract (median OS = 66 months, 5-year OS = 55%, log-rank test P = .001, Fig. 2A) or head and neck (median OS = 48 months, 5-year OS = 63%, log-rank test P < .001, Fig. 2B). Patients with Heme/RES disease had similar survival to those with lymph node disease (5-year OS = 41%, log-rank test P = .84, Fig. 2C).

Figure 2.

The Kaplan-Meier survival curves are shown for patients with mantle cell lymphoma. Each figure compares the survival curves for patients with an extranodal primary site of disease to cases with primary site of lymph node: (A) gastrointestinal (GI) tract, (B) head and neck, and (C) hematologic and reticuloendothelial system (Heme/RES).

Univariate Cox regression models showed that MIPI age categories (age > 69 years HR = 4.1, 95% CI = 3.3-5.0), advanced stage (HR = 1.4, 95% CI = 1.3-1.6) and presence of B symptoms at diagnosis (HR = 1.6, 95% CI = 1.4-1.8) were predictors of worse survival. When compared with lymph node primary sites, presence of GI tract (HR = 0.8, 95% CI = 0.6-0.9) and head and neck (HR = 0.6, 95% CI = 0.5-0.7) primary sites predicted better survival (Table 3). Other extranodal primary sites did not predict better survival.

Table 3. Univariate and Multivariate Cox Proportional Regression Models for Predictors of Survival
 Univariate ModelMultivariate Model
VariablesHR95% CIHR95% CI
  1. Abbreviations: CI, confidence interval; Heme/RES, hematologic and reticuloendothelial system; HR, hazard ratio.

 Asian/Pacific Islander0.990.79-
Age category, y    
 Localized (Stage I/II)1Reference1Reference
 Advanced (Stage III/IV)1.421.26-1.591.391.22-1.59
Diagnostic era    
 Before 20051Reference1Reference
 After 20040.930.85-1.030.890.78-0.98
B symptoms    
Primary site    
 Lymph nodes1Reference1Reference
 Gastrointestinal tract0.750.63-0.890.750.62-0.91
 Head and neck0.590.48-0.730.680.55-0.85
Disease extension    
 Nodal disease1Reference1Reference
 Extranodal disease0.740.66-0.83
Surgical treatment    
 Did not receive1Reference1Reference
Radiation treatment    
 Did not receive1Reference1Reference

In a multivariate model that included 4100 cases, female sex (HR = 0.9, 95% CI = 0.8-0.9, Table 3), primary disease of the GI tract (HR = 0.8, 95% CI = 0.6-0.9) and of head and neck (HR = 0.7, 95% CI = 0.6-0.9) predicted better survival, whereas black race (HR = 1.4, 95% CI = 1.1-1.7), MIPI age categories (age > 69 HR = 4.3, 95% CI = 3.5-5.4), and advanced stage (HR = 1.4, 95% CI = 1.2-1.6) predicted worse survival. Presence of B symptoms was excluded as a variable in the multivariate model, because 33.5% of patients had missing data for this variable. Because primary site was significantly associated with the presence of B symptoms (Table 2), a second model including the B symptom variable was constructed, despite the loss of observations. In this model, extranodal primary site remained statistically significant as a predictor for better survival (data not shown). Tests of interaction between stage and primary site were not significant.


To our knowledge, this is the first population-based study that evaluates the association between primary site of disease and survival in patients with MCL. Clinically, MCL commonly presents extranodally,27 particularly in the GI tract where involvement may be subtly detected on biopsy28, 29 or extensively as in the case of multiple lymphomatous polyposis.29-32 Two studies independently estimated that 80% to 90% of patients with MCL had GI tract involvement.28, 29 Among those with GI involvement, 8% did not have concurrent disease of the Heme/RES, consistent with the proportion of patients with primary GI disease in our cohort (7.8%). Similarly, the literature reports frequent peripheral blood involvement, ranging from low concentrations of “spill over” tumor cells in patients with nodal disease to patients with a leukemic presentation, the definitions of which vary.16, 17, 33, 34 In our analysis, 3.6% of patients presented with a Heme/RES primary site, attributed most commonly to the bone marrow (53%) or spleen (46%).

MCL is thought to have both aggressive and indolent phenotypes, but identifying patients with indolent disease remains difficult. Several research groups have defined indolence as stable disease without the need for treatment over a variable period of time,16, 34, 35 a definition which also identifies a subset of patient with favorable MIPI scores.35-37 In our cohort, patients with primary disease of the GI tract or head and neck were more likely to present with localized disease, without B symptoms, and to have longer OS than the reference group. The lesser extent of disease in these patients may result from earlier detection through routine examinations and procedures such as colonoscopies or through the manifestation of symptoms at these sites; however, differences in outcome may reflect distinctions in the biology of disease at each site. A precedent for a lymphoma primary site acting as a surrogate for tumor biology is found in the case of primary cutaneous B cell lymphoma, leg-type (PCBCL-LT). PCBCL-LT was first distinguished from PLBCL at other skin sites for its occurrence in older patients, worse 5-year survival rates, and its characteristic presentation on the lower limbs. Its distinction in classification schemes facilitated the discovery of histological and genetic differences that confirmed its uniqueness from other subtypes.38-41 In the case of MCL, primary disease of the GI tract and the head and neck was more common in Asians than in either whites or blacks, perhaps alluding to the roles of genetic, dietary, or other environmental factors in the development of MCL.42-46

In addition to behavioral and biological differences, OS may be influenced by earlier detection or differences in treatment between groups. Patients with head and neck primaries were more likely to receive radiation or surgery. More specific treatment data might reveal even larger differences in treatment strategies. After controlling for age, disease stage, and treatment modality in a multivariate model, patients with primary disease of the GI tract and head and neck still had superior survival, arguing for its role as a prognostic factor, and perhaps, an indicator of indolent behavior. This analysis is limited by the lack of data on 3 of the 4 MIPI criteria and more-specific treatment data. Inclusion of these data may render primary site insignificant as a predictor of survival, highlighting the need for cohorts with more detailed clinical information and a greater capacity to handle lead-time bias to delineate whether these differences in OS arise primarily from care associated with these patterns of presentation or from other biological factors. These findings may also be valuable in identifying superior management strategies.

Several studies have found associations between non-nodal disease and survival.15-18 Non-nodal disease is often used interchangeably with leukemic MCL because most cases of non-nodal disease have evidence of peripheral blood, bone marrow, or splenic involvement. Our study also found an improvement in survival among patients with primary Heme/RES disease, but it was not statistically significant. Discrepancies in the significance of this relationship probably stem from differences in the classification of leukemic disease and primary Heme/RES disease.

These studies have also correlated non-nodal, leukemic disease to other potential biomarkers of indolent disease. Orchard et al found that 44% of patients with non-nodal disease lacked somatic hypermutation of the immunoglobulin heavy chain (IgVH), a biomarker of B cell origin and a prognostic factor in chronic lymphocytic leukemia, compared with 90% in those with nodal disease.15 Fernàndez et al found that indolence was associated with non-nodal presentation, lower MIPI score, IgVH hypermutation, noncomplex karyotypes, and the lack of expression of 13 genes, including SOX11 (SRY [sex determining region Y]-box 11), that were all expressed in conventional MCL.16 Ondrejka et al found a similar profile in patients with indolent disease,17 whereas Nygren et al found no differences in SOX11 expression.34 Prospective, population-based observational studies that capture detailed clinical information on primary site, prognostic factors, laboratory variables, treatment, and treatment outcomes are needed to discern the role that primary site plays when these other factors are measured. Such studies should also collect biological samples at diagnosis to determine the associations between site and biomarkers for improved survival. Our group has recently performed a similar study for patients with diffuse large B cell lymphoma,47 and population-based studies for MCL and other NHLs are planned.

Similar to all other SEER studies, lack of central pathology review was a limitation of our study. The determination of disease classification and severity is dependent on local diagnostic practices and standards, which may vary.48-51 Nevertheless, pathologist agreement in diagnosing MCL is high when histology, immunophenotype, and clinical data are used.3 Another possible limitation of this study is the use of data classified using ICD-O-2, before the introduction of the revised WHO classification in 2001. However, Clarke et al showed an 81% agreement between computer-converted ICD-O-2 codes to ICD-O-3 codes and registry-assigned codes for MCL cases diagnosed between the 1998 and 2000 SEER.52


In patients with MCL, primary site may serve as a tool for identifying patients with indolent disease and ultimately help in guiding clinical management. In our analysis, patients with primary disease of the GI tract and the head and neck had better risk profiles and superior survival compared with patients with primarily nodal disease. Primary site may correlate with certain biological characteristics associated with disease behavior and pathogenesis, but additional prospective cohort studies are needed. Asians also had a significantly higher proportion of extranodal disease as compared with whites and blacks, suggesting that disease site may be influenced by genetic or environmental factors. Future studies should examine the biological underpinnings of indolent disease for individuals presenting with GI and head and neck primary sites among populations of patients with MCL.


This work was supported by a Georgia Cancer Coalition Distinguished Scientist Award reduce and Cancer Research Award, R21 CA158686-01A1, and by an American Society of Hematology Amos Medical Faculty Development Award to Dr. Flowers.


Dr. Flowers has been a consultant for Celgene, Spectrum, OptumRx, Seattle Genetics, Genentech/Biogen-Idec/Roche (unpaid), Millennium/Takeda (unpaid) and has received research funding from Millennium/Takeda, Novartis, Gilead, and Spectrum.