1612 Treatment, Patterns of Failure, and Survival of Patients with Stage I Nodal and Extranodal NonHodgkin’s Lymphomas, According to Data in the Population-Based Registry of the Comprehensive Cancer Centre West Augustinus D. Krol, M.D.1 Jo Hermans, Ph.D.2 Lilian Dawson, M.D.3 Simone Snijder, B.Sc.4 Pierre W. Wijermans, M.D., Ph.D.5 Hanneke C. Kluin-Nelemans, M.D., Philip M. Kluin, M.D., Ph.D.6 J. Han van Krieken, M.D., Ph.D.6 Evert M. Noordijk, M.D., Ph.D. 7 3 Ph.D. 1 Department of Radiotherapy, University Hospital Rotterdam/Daniel den Hoed Cancer Center, Rotterdam, the Netherlands. 2 Department of Medical Statistics, University of Leiden, Leiden, the Netherlands. 3 Department of Hematology, Leiden University Hospital, Leiden, the Netherlands. 4 The Comprehensive Cancer Center West, Leiden, the Netherlands. 5 Department of Hematology, Leyenburg Hospital, Den Haag, the Netherlands. 6 Department of Pathology, Leiden University Hospital, Leiden, the Netherlands. 7 Department of Clinical Oncology, Leiden University Hospital, Leiden, the Netherlands. BACKGROUND. Primary extranodal lymphomas (EN-NHLs) are a heterogeneous category of tumors that are considered to be different from primary nodal nonHodgkin’s lymphomas (N-NHLs). To what extent these differences have clinical implications is currently not very clear, because knowledge of EN-NHL as a separate group is limited. METHODS. Using data from the Comprehensive Cancer Centre West (CCCW) population-based NHL registry in the Netherlands, N-NHL and EN-NHL patients were compared to determine differences in characteristics at diagnosis, responses to treatment, patterns of failure, and survival. RESULTS. At presentation, EN-NHL patients had poorer performance scores and more often bulky tumors compared with N-NHL patients, resulting in poorer responses to treatment (complete response rates were 72% and 84%, respectively; P 5 0.04) and inferior 5-year overall survival (49% and 63%, respectively; P 5 0.003). Among EN-NHL patients, considerable variations in response, survival, and relapse rates were observed, with gastric NHL patients having the best and central nervous system NHL patients having the worst prognosis (66% and 7% 5-year overall survival, respectively). Relapse rates for N-NHL and EN-NHL patients did not differ (39% and 36% 5-year relapse rates, respectively), whereas among ENNHL patients considerable differences in relapse rates were noted. Relapses among N-NHL patients were mainly found in nodal sites, whereas recurrent disease in EN-NHL patients was mainly found in extranodal sites. CONCLUSIONS. In this population-based study, Stage I EN-NHL patients as a group had a poorer prognosis than N-NHL patients. However, among EN-NHL patients, considerable differences in response, relapse risk, and survival were observed. The failure analysis conducted in this study suggests that patterns of dissemination for N-NHL and EN-NHL are different. Cancer 1998;83:1612–9. © 1998 American Cancer Society. KEYWORDS: non-Hodgkin’s lymphoma, nodal, extranodal, prognosis, patterns of dissemination. Address for reprints: Augustinus D. Krol, M.D., Department of Radiotherapy, University Hospital Rotterdam/Daniel den Hoed Cancer Center, P.O. Box 5201, 3008 AE Rotterdam, the Netherlands. Received December 26, 1997; revision received April 6, 1998; accepted April 6, 1998. © 1998 American Cancer Society U p to 40% of non-Hodgkin’s lymphomas (NHLs) originate in sites outside the lymph nodes.1– 4 These primary extranodal lymphomas (EN-NHL) are a heterogeneous category of tumors that are considered to be different from primary nodal non-Hodgkin’s lymphomas (N-NHL) with respect to patterns of dissemination and oncogene expression.6 To what extent these differences have clinical implica- Stage I Nodal and Extranodal NHL/Krol et al. 1613 TABLE 1 Disease Localization and Malignancy Grade (WF) of all Stage I NHL Patients (n 5 264) in the CCCW Population-Based Registry No. of patients Total L I H U Nodal Lymph node Waldeyer’s tonsillar ring Spleen Extranodal Head/necka Stomach Intestine Bone Connective tissue Skin Female breast Testis Bladder Orbita/lacrimal gland Brain Spinal cord, epi-/extradural Thyroid 114 96 17 1 150 19 47 17 7 6 11 5 7 1 6 14 5 5 37 35 1 1 14 2 4 2 – – 1 – – – 1 2 – 2 59 45 14 – 99 12 38 9 5 4 7 4 6 1 2 5 3 3 11 10 1 – 18 – 3 5 2 – 1 – – – 1 6 – – 7 6 1 – 19 5 2 1 – 2 2 1 1 – 2 1 2 – Total 264 51 158 29 26 NHL: non-Hodgkin’s lymphoma; CCCW: Comprehensive Cancer Centre West; WF: Working Formulation; L: low grade; I: intermediate grade; H: high grade; U: unclassifiable. a Waldeyer’s tonsillar ring was excluded. tions is currently not very clear because knowledge of EN-NHL as a separate category is limited. Data on the clinical behavior of EN-NHL have come mainly from hospital-based series, in which bias due to patient selection was unavoidable. Therefore, it remains unclear whether the picture of EN-NHL that has emerged from these studies reflects its true nature. Populationbased information about clinical behavior of EN-NHL is scarce.2,3 In the current study, we made a population-based comparison of N-NHL and EN-NHL and analyzed possible differences in patient characteristics at diagnosis, responses to treatment, and patterns of failure and survival, with the aim of determining whether extranodal lymphomas have specific clinical behavior. To avoid uncertainties about primary nodal or extranodal origin of lymphomas, this study was limited to patients with a single affected lymph node or extranodal organ (Ann Arbor Stages I and IE). PATIENTS AND METHODS All hospitals (15) and pathology laboratories (9) in the western part of the Netherlands covered by the Comprehensive Cancer Centre West (CCCW)(1.6 million inhabitants) cooperated in this study. From June 1981 to December 1989, data on all newly diagnosed NHL patients (n 5 1168) living in the region were entered into a registry by the CCCW Hemato-Oncology Group. Patients with primary cutaneus T-cell lymphoma, plasmacytoma, acute lymphoblastic leukemia, and chronic lymphocytic leukemia and all cases diagnosed at postmortem examination were excluded. Diagnosis and classification by the local pathologist were reviewed by a panel of pathologists. Lymphomas were classified according to the original Kiel classification and graded according to the malignancy grades of the Working Formulation for Clinical Usage (WF).7 Cases included in the current study were reclassified according to the revised European–American classification of lymphoid neoplasms (REAL)8 and grouped into the categories “indolent,” “aggressive,” and “very aggressive” proposed by Hiddeman et al.9 Complete clinical evaluation included history, physical examination, performance status, leukocyte count and differential, lactate dehydrogenase (LDH), chest radiograph, computed tomography (CT) scan of the abdomen or lymphangiogram supplemented by isotope spleen and liver scan in the early years of the registration, morphologic examination of bone marrow aspirate and biopsy, and biopsy of involved tissues. Clinical parameters and involved lymph nodes and extranodal sites were documented. Patients were staged according to the 1614 CANCER October 15, 1998 / Volume 83 / Number 8 Ann Arbor classification.10 Disease localizations in the lymph nodes, Waldeyer’s tonsillar ring, and the spleen were considered to be N-NHL, and other localizations were considered to be EN-NHL. Patients were treated according to the preference of the local physician. Response to treatment was defined according to standard criteria, with complete response (CR) defined as the disappearance of all clinical, radiologic, or other evidence of disease. For patients who experienced disease recurrence after any CR, involved sites and nodal and/or extranodal status of recurrence were recorded, using the same definitions as for the primary disease localization. Recurrence was defined as local when the initially affected organ or first-echelon lymph nodes were involved and as distant if other sites were involved. As some patients had both local and distant failures, the total number of recurrences was greater than the number of patients with recurrent disease. Overall survival (OS) was calculated from diagnosis until death (of all causes) or last follow-up (1996). Recurrence free survival (RFS) was calculated for patients in CR only, from the date of CR until recurrence (failure) or last follow-up (including death in CR). Statistical analysis was performed using chisquare tests to compare percentages in cross-tabulations and the log rank test to compare Kaplan–Meier survival curves. TABLE 2 Patient Characteristics of All Stage I NHL Patients (n 5 264) in the CCCW Population-Based Registry Characteristics Gender Male Age (yrs) $60 Performance (KPS)a ,80 LDHb .Normal Tumor bulkc $5 cm Complete staging No Grade (WF) Low Intermediate High Unclassifiable REAL classification Indolent Aggressive Very aggressive Unclassifiable Nodal (n 5 114) % Extranodal (n 5 150) % P value 50 46 ns 57 63 ns 7 32 ,0.001 21 22 ns 28 46 0.009 27 31 ns 33 52 10 6 9 66 12 13 ,0.001 37 52 3 8 19 72 3 6 0.005 NHL: non-Hodgkin’s lymphoma; CCCW: Comprehensive Cancer Centre West; KPS: Karnofsky performance status; LDH: lactate dehydrogenase; WF: Working Formulation; REAL: revised European– American classification of lymphoid neoplasms; ns: not significant. Missing: a 15%, b 16%, c 30%. RESULTS Patients Of 1168 patients in the CCCW database, 264 (23%) were considered Ann Arbor Stage I. Seventy-seven patients (29%), most of them of advanced age and/or with poor performance scores, were not completely staged, usually because no bone marrow biopsy was performed. As differences in response, OS, and RFS between N-NHL and EN-NHL patients were not dependent on the completeness of staging, we decided to include all 264 Stage I patients (completely and incompletely staged) in our study. One hundred fourteen patients (43%) presented with N-NHL and 150 (57%) with EN-NHL. In patients with N-NHL, the lymph nodes of the head and neck were most frequently involved. The majority of ENNHLs were localized in the gastrointestinal tract, but localizations in a wide diversity of organs were observed (Table 1). Patient characteristics are presented in Table 2. EN-NHL patients more often presented with bulky disease (especially patients with gastric and intestinal localizations) and poor performance status (especially patients with gastric and central nervous system [CNS] localizations) than N-NHL patients. Ex- tranodal lymphomas were more often of intermediate and high grade malignancy (WF) than nodal lymphomas (Table 2, P , 0.001). Pathologic classification (REAL) of all lymphomas is presented in Table 3. Most EN-NHLs were classified as diffuse large B-cell and mucosa-associated lymphoid tissue (MALT) high grade lymphomas (according to Isaacson), whereas the majority of N-NHLs were classified as follicular lymphomas or diffuse large B-cell lymphomas. As the relation between Helicobacter pylori infection and MALT lymphomas was not known during the registration period, there were no data on test results for Helicobacter pylori infection in patients with MALT lymphomas. Information on the incidence of infection with the human immunodeficiency virus (HIV) in this series of patients would have given noteworthy additional information, because HIV infection is known to be associated with high grade, often extranodal lymphomas. Due to the finding that HIV infections were rare in the CCCW region during the registration period, tests for HIV antibodies were not performed on a regular basis, but we expected the number of HIVrelated lymphomas in this cohort to be negligible. Stage I Nodal and Extranodal NHL/Krol et al. 1615 TABLE 3 Histologic Classification (REAL) of All Stage I NHL Patients (n 5 264) in the CCCW Population-Based Registry No. of patientsa Nodal Extranodal Total Indolent lymphomas (low risk) B-cell chronic lymphocytic leukemia/lymphoma Lymphoplasmacytoid lymphoma Follicle center cell lymphoma, follicular Follicle center cell lymphoma, diffuse Marginal zone lymphoma, MALT low grade Aggressive lymphomas (intermediate risk) Mantle cell lymphoma Follicle center cell lymphoma, large cell Diffuse large B-cell lymphoma Diffuse large B-cell lymphoma, MALT high grade peripheral T-cell lymphoma, unspecified Very aggressive lymphomas (high risk) Burkitt-like lymphoma Precursor B-lymphoblastic lymphoma/leukemia Unclassifiable 42 (29) 2 (1) – 36 (25) 4 (3) – 59 (46) 4 (4) 9 (8) 42 (31) – 3 (3) 3 (3) 2 (2) 1 (1) 10 (5) 28 (17) 1 (1) 5 (3) 5 (3) 1 (1) 16 (9) 108 (76) 1 (1) 1 (1) 55 (40) 45 (33) 6 (2) 5 (4) 4 (3) 1 (1) 9 (7) 70 (46) 3 (2) 5 (3) 41 (28) 5 (4) 16 (9) 167 (122) 5 (5) 10 (8) 97 (71) 45 (33) 10 (5) 8 (7) 6 (5) 2 (2) 19 (11) Total 114 (83) 150 (103) 264 (187) NHL: non-Hodgkin’s lymphoma; CCCW: Comprehensive Cancer Centre West; REAL: revised European–American classification of lymphoid neoplasms; MALT: mucosa-associated lymphoid tissue. a Nos. in parentheses represent patients who were completely staged (n 5 187). Treatment Local treatment (surgery and/or radiotherapy) was given to 180 patients (68%) and systemic treatment (chemotherapy alone or in combination with surgery and/or radiotherapy) to 59 (22%). Twenty-five patients (10%) did not receive any treatment, in most cases because of advanced age and/or poor performance status. Surgery was performed twice as often on ENNHL patients (64%) than on N-NHL patients (32%). No differences were observed between N-NHL and EN-NHL patients regarding the application of chemotherapy and radiotherapy or the decision to refrain from treatment. Responses One hundred eighty-seven (78%) of 239 patients who received treatment achieved a CR during the course of their disease. EN-NHL patients had a significantly lower CR rate (72%) than N-NHL patients (84%) (P 5 0.04). For both N-NHL and EN-NHL patients, age 60 years (P 5 0.05), Karnofsky performance status $80 (P , 0.001), normal lactate dehydrogenase (LDH) level (P , 0.001), and tumor bulk , 5 cm (P 5 0.004) were of favorable prognostic significance for CR, whereas malignancy grade (WF) was not. EN-NHL patients younger than 60 years with elevated LDH or bulky tumors had inferior CR rates than N-NHL patients with similar characteristics (Table 4). Response rates for the different nodal and extranodal primary sites are summarized in Table 5. The inferior response rate for EN-NHL patients was mainly a result of the poor CR rates of patients with intestinal and CNS NHL. Overall Survival OS for the entire patient cohort, including patients who did not receive any treatment, was 55% at 5 years. The OS for EN-NHL patients was inferior to the OS for N-NHL patients (49% and 63% at 5 years, respectively; P 5 0.003). Age 60 years (P , 0.001), performance status $80 (P , 0.001), normal LDH level (P , 0.001), tumor bulk ,5 cm (P 5 0.006), and low grade malignancy (WF) (P 5 0.005) were of favorable prognostic significance regarding OS for both N-NHL and ENNHL patients. A comparison of the prognostic factors affecting the OS of N-NHL and EN-NHL patients is presented in Table 4. EN-NHL patients younger than 60 years with elevated LDH or bulky tumors had lower OS at 5 years than N-NHL patients with similar characteristics. OS data in relation to involved site at diagnosis are presented in Table 5. The difference in the OS for N-NHL and EN-NHL patients was mainly a resulted of the poor survival of patients with intestinal, brain, and female breast localizations. Patients with a stomach localization fared better than other 1616 CANCER October 15, 1998 / Volume 83 / Number 8 TABLE 4 Prognostic Factors for Complete Response, Overall Survival, and Relapse Risk for Stage I Nodal and Extranodal NHL Patients in the CCCW Population-based Registrya % CR Total Completely staged patients Age (yrs) ,60 $60 KPS $80 ,80 LDH Normal .Normal Tumor bulk ,5 cm $5 cm Grade (WF) Low Intermediate High 5-yr OS 5-yr RR Nodal EN P value Nodal EN P value Nodal EN P value 84 86 72 75 0.04 0.007 63 68 49 52 0.003 0.009 39 27 36 33 ns ns 91 78 75 70 0.02 ns 85 46 66 39 0.02 ns 30 48 32 39 ns ns 84 67 81 51 ns ns 67 29 57 31 ns ns 39 62 40 38 ns ns 87 81 86 44 ns 0.02 67 57 56 20 ns ,0.001 39 38 39 55 ns ns 84 89 89 61 ns 0.01 64 62 63 40 ns 0.02 32 58 36 44 ns ns 88 83 73 85 74 65 ns ns ns 75 56 54 71 52 33 ns ns ns 29 38 50 18 29 49 ns ns ns NHL: non-Hodgkin’s lymphoma; CCCW: Comprehensive Cancer Centre West; CR: complete response; OS: overall survival; RR: relapse risk; EN: extranodal; KPS: Karnofsky performance status; LDH: lactate dehydrogenase; WF: Working Formulation; ns: not significant. a OS data on all patients (n 5 264), CR data on patients who received treatment (n 5 239), RR data on patients in CR (n 5 187). EN-NHL patients and had OS comparable to that of N-NHL patients. Relapse Risk The 5-year relapse risk for all patients was 37%. Of all factors tested, including completeness of staging, only malignancy grade (WF) was of prognostic significance for relapse. Relapse rates for low, intermediate, and high grade lymphomas were 26%, 33%, and 55% at 5 years, respectively (P 5 0.01). By comparison, N-NHL and EN-NHL patients had similar relapse risks at 5 years, also when compared within subgroups of the prognostic factors identified for CR and OS (Table 4). Relapse risk within the group of patients with extranodal lymphoma differed markedly (Table 5). Patients with gastric lymphomas had a significantly lower relapse risk (16% at 5 years) than patients with other extranodal or nodal lymphomas (36% and 39% relapse risk at 5 years, respectively; P 5 0.007). Patterns of Failure Risk for local and distant relapse at 5 years was 25% and 22%, respectively. Elevated LDH level (P 5 0.002) and tumor bulk $5 cm (P 5 0.01) were prognostic for local relapse, and for distant relapse risk the same trend was observed. Local and distant relapse rates at 5 years for N-NHL and EN-NHL patients did not differ. However, patients who presented with nodal disease experienced failure predominantly in nodal localizations, whereas the majority of those presenting with extranodal disease had extranodal recurrences; this difference in preferential relapse sites between N-NHL and EN-NHL patients was significant (P 5 0.04). We looked for specific patterns of failure in subgroups of extranodal lymphomas, but the number of recurrences in these groups was too small to permit conclusions. Some noteworthy trends were observed, however. In the group of MALT lymphoma patients, approximately half of the recurrences were localized in the digestive and/or respiratory tract, whereas gastric recurrence was observed in two patients with a Waldeyer’s tonsillar ring localization. All recurrences of patients with cutaneous lymphomas occurred in the skin, and all patients with CNS NHL experienced local failure. An analysis of patterns of failure for individual nodal and extranodal primary sites is presented in Table 6. DISCUSSION This study of Stage I NHL patients was performed to investigate possible differences in the clinical picture for patients with primary nodal and extranodal Stage I Nodal and Extranodal NHL/Krol et al. TABLE 5 Complete Response, Overall Survival, and Relapse Risk for Stage I NHL Patients in the CCCW Population-Based Registry, Analyzed by Primary Disease Localizationa Nodal (114) Lymph node (96) Waldeyer’s ring (17) Extranodal (150) Head/neck6 (19) Stomach (47) Intestine (17) Bone/connective tissue (13) Skin (11) Female breast (5) Testis (7) Orbita/lacrimal gland (6) Brain (14) Spinal cord, epi-/extradural (5) Thyroid (5) CR % 5-yr OS (%) 5-yr RR (%) 84 84 88 72 72 80 40 60 78 100 100 100 43 80 80 63 66 47 49 53 66 35 50 45 20 43 83 7 57 40 39 38 43 36 31 16 33 33 43 100 29 72 50 62 50 NHL: non-Hodgkin’s lymphoma; CCCW: Comprehensive Cancer Centre West; CR: complete response; OS: overall survival; RR: relapse risk. a OS data on all patients (n 5 264), CR data on patients who received treatment (n 5 239), RR data on patients in CR (n 5 187). b Waldeyer’s tonsillar ring was excluded. disease. By analyzing patients from a populationbased registry, we circumvented the problem of bias due to patient selection. The absence of selection is reflected in the advanced age and poor performance status of the patients in this study, which resulted in a decision to refrain from treating 10% of the patients. When this is considered, it is not surprising that these treatment results were inferior to those for published hospital-based series, which usually report on a younger selection of patients with aggressive NHL.11–15 The goal of this study was to compare primary nodal and extranodal lymphoma patients with respect to their clinical situations at diagnosis, treatment, responses to treatment, relapse risk, patterns of failure, and survival. The percentage of EN-NHL patients in this series was high in comparison with other population-based studies,1,3,4 probably due in part to the fact that only Stage I patients were selected. The relative frequency of extranodal organs affected, with gastrointestinal tract and localizations in the head and neck area the most commonly involved sites, was in line with other populationbased data.1,3,4 EN-NHL patients had relatively low performance scores, bulky tumors, and more often lymphomas of intermediate or high grade malignancy at diagnosis. Given the finding that the treatment of 1617 N-NHL and EN-NHL patients (including the decision to refrain from treatment) was not essentially different, the reason for the observed inferior response rate and survival of EN-NHL patients must be a result of the poor prognostic characteristics of these patients at diagnosis. Patients with primary EN-NHL and a considerable tumor load (elevated LDH and bulky disease) or intermediate and high grade lymphomas appeared to have especially poor prognoses compared with N-NHL patients with the same characteristics. The relatively poor prognostic characteristics of EN-NHL patients became irrelevant, however, when CR was achieved following treatment: in these cases, recurrence free survival for N-NHL and EN-NHL patients was equal. When the differences in prognostic factors between N-NHL and EN-NHL were analyzed in more detail, the poor prognostic factors for EN-NHL patients were not characteristic of the entire group of these patients; rather, they were observed in certain subgroups of patients. Poor performance status was mainly observed in patients with gastric and central nervous system localizations. Bulky disease was mainly found in patients with gastric, intestinal, and bone or soft tissue localizations. As a consequence, treatment results and survival for these patients were inferior to those for other EN-NHL patients. However, patients with a gastric localization were an exception to this rule: despite poor performance and considerable tumor load at diagnosis, the survival of these patients was superior to that of other EN-NHL patients and comparable to that of N-NHL patients. Our analysis demonstrated that this was a result of the relatively low relapse risk for gastric lymphoma patients. As 45 of 47 gastric lymphomas were classified as MALT lymphomas of either low or high malignancy grade, this observation supports the MALT concept formulated by Isaacson.16 This concept postulated that MALT lymphomas had restricted dissemination due to “homing” receptors on the cell membranes of MALT lymphoma cells, resulting in a favorable prognosis for patients with these lymphomas. In an analysis of all primary gastric lymphomas in the CCCW database (disseminated cases included), the clinical relevance of this concept was also illustrated.17 This study also focused on possible specific patterns of dissemination for extranodal lymphomas other than MALT lymphomas. Although the relapse risk for N-NHL and EN-NHL patients was similar, also with respect to the distribution of local and distant failures, a striking difference between these patients was seen with respect to preferential sites for recurrence. N-NHL patients mainly experienced recurrence in nodal sites, and EN-NHL patients in extranodal 1618 CANCER October 15, 1998 / Volume 83 / Number 8 TABLE 6 Patterns of Failure for Stage I NHL Patients in the CCCW Population-Based Registry, in Complete Response after Treatment (n 5 187) No. of patients CR Failure free Local failure Distant failure Local 1 distant failure Nodal (114) Lymph node (96) Waldeyer’s tonsillar ring (17) Spleen (1) Extranodal (150) Head/necka (19) Stomach (47) Intestine (17) Bone/connective tissue (13) Skin (11) Female breast (5) Testis (7) Orbita/lacrimal gland (6) Brain (14) Spinal cord, epi-/extradural (5) Thyroid (5) 88 73 14 1 99 13 32 6 9 7 5 7 6 6 4 4 55 46 9 – 64 10 27 3 5 4 – 6 2 3 2 2 13 9 3 1 13 – 1 1 1 1 2 – 3 3 – 1 9 9 – – 14 2 3 2 3 – 1 1 1 – 1 – 11 9 2 – 8 1 1 – – 2 2 – – – 1 1 Total 187 118 26 23 19 NHL: non-Hodgkin’s lymphoma; CCCW: Comprehensive Cancer Centre West; CR: patients in complete response after treatment. a Waldeyer’s tonsillar ring was excluded. sites. This observation supports the concept of “directed migration” in the dissemination of NHL, regulated by adhesion molecules on the cellular membranes of lymphoma cells.5 In conclusion, this study demonstrated that primary extranodal NHL patients as a group have a poorer response to treatment and consequently survival inferior to that of patients with primary nodal NHL. However, among EN-NHL patients, considerable variations in response to treatment and survival exist, with gastric lymphoma patients having the best and central nervous system lymphoma patients having the worst prognosis. Consequently, for clinical purposes, it seems more useful to make distinctions based on specific sites, tumor grade, and histologic classification than to consider ENNHL a single entity. In cases of CR to treatment, differences between nodal and extranodal lymphoma patients disappeared: no differences in relapse rates were observed. However, patterns of relapse for nodal and extranodal lymphomas were different, suggesting that sites were preferential for recurrence depending on the primary site. In light of this, it also seems more important to consider the primary site rather than the nodal or extranodal status of NHL. REFERENCES 1. 2. Freeman C, Berg JW, Cutler SJ. Occurrence and prognosis of extranodal lymphomas. Cancer 1972;29:252– 60. Otter R, Gerrits WBJ, vd Sandt MM, Hermans J, Willemze R. Primary extranodal and nodal non-Hodgkin’s lymphoma: a survey of a population-based registry. Eur J Cancer Clin Oncol 1989;25:1203–10. 3. d’Amore F, Christensen BE, Brincker H, Pedersen NT, Thorling K, Hastrup J, et al. Clinicopathological features and prognostic factors in extranodal non-Hodgkin’s lymphomas. Eur J Cancer 1991;27:1201– 8. 4. Newton R, Ferlay J, Beral V, Devesa S. The epidemiology of non-Hodgkin’s lymphoma: comparison of nodal and extranodal sites. Int J Cancer 1997;72: 923–30. 5. Pals ST, Drillenburg P, Radaskiewicz T, Manten-Horst E. Adhesion molecules in the dissemination of non-Hodgkin’s lymphomas. Acta Haematol 1997;97:73– 80. 6. Raghoebier S, Kramer MHH, van Krieken JHJM, de Jong D, Limpens J, Kluin-Nelemans JC, et al. Essential differences in oncogene involvement between primary nodal and extranodal large cell lymphoma. Blood 1991; 78:2680 –5. 7. Non-Hodgkin’s lymphoma pathological classification project. National Cancer Institute sponsored study of classifications of non-Hodgkin’s lymphomas. Summary and descriptions of a Working Formulation for clinical use. Cancer 1982;49,2112–35. 8. Harris NL, Jaffe ES, Stein H, Banks PM, Chan JKC, Cleary ML, et al. A revised European American Classification of lymphoid neoplasms: a proposal from the International Lymphoma Study Group. Blood 1994;84:1361. 9. Hiddeman W, Longo DL, Coiffier B, Fisher RI, Cabanillas F, Cavalli F, et al. Lymphoma classification: the gap between biology and clinical management is closing. Blood 1996;88: 4085–9. 10. Carbone PP, Kaplan HS, Musshoff K, Smithers DW, Tubiana M. Report of the committee on Hodgkin’s disease staging classification. Cancer Res 1971;31:1860 –1. Stage I Nodal and Extranodal NHL/Krol et al. 11. Prestidge BR, Horning SJ, Hoppe RT. Combined modality therapy for stage I–II large cell lymphoma. Int J Radiat Oncol Biol Phys 1988;15:633–9. 12. Jones S, Miller TP, Connors JM. Long term follow-up and analysis for prognostic factors for patients with limited stage diffuse large cell lymphoma treated with initial chemotherapy with or without adjuvant radiotherapy. J Clin Oncol 1989;9:1186 –91. 13. Connors JM, Klimo P, Fairey RN, Voss N. Brief chemotherapy and involved field radiation therapy for limited stage, histologically aggressive lymphoma. Ann Intern Med 1987; 107:25–30. 14. Velasquez WS, Fuller LM, Jagannath S, Tucker SL, North LB, Hagemeister FB, et al. Stages I and II diffuse large cell 1619 lymphomas: prognostic factors and long term results with CHOP-Bleo and radiotherapy. Blood 1991;77:942–7. 15. Tondini C, Zanini M, Lombardi F, Bengala C, Rocca A, Giardini R, et al. Combined treatment with primary CHOP chemotherapy followed by locoregional irradiation in stage I and II histologically aggressive non-Hodgkin’s lymphomas. J Clin Oncol 1993;11:720 –5. 16. Isaacson PG. The MALT lymphoma concept updated. Ann Oncol 1995;6:319 –20. 17. Krol ADG, Hermans J, Kramer MHH, Kluin PhM, KluinNelemans JC, Blok P, et al. Gastric lymphomas compared with lymph node lymphomas in a population-based registry differ in stage distribution and dissemination patterns but not in patient survival. Cancer 1997; 79:390 –7.