Primary extranodal non-Hodgkin’s lymphoma (NHL): the impact of alternative definitions tested in the Comprehensive Cancer Centre West population-based NHL registry

A. D. G. Krol1,+, S. le Cessie2, S. Snijder3, J. C. Kluin-Nelemans4, P. M. Kluin5 and E. M. Noordijk1

Departments of 1 Clinical Oncology and 2 Medical Statistics, Leiden University Medical Centre, Leiden; 3 The Comprehensive Cancer Centre West, Leiden; Departments of 4 Hematology and 5 Pathology, University Hospital Groningen, Groningen, The Netherlands

Received 12 July 2002; accepted 23 July 2002


    Abstract
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
Background:

The definition of primary extranodal non-Hodgkin’s lymphoma (NHL) is a controversial issue, especially in patients where both nodal and extranodal sites are involved.

Patients and methods:

The impact of different definitions of primary extranodal NHL on incidence and prognosis is explored using data from a population-based NHL registry.

Results:

Using liberal criteria, 389 (34%) cases were classified as primary extranodal NHL. Overall survival (OS) rates of nodal and extranodal NHL patients defined this way were comparable; however, extranodal NHL patients had a better disease-free survival (DFS). When strict criteria were applied, 231 cases (20%) were classified as primary extranodal NHL. OS and DFS rates of extranodal NHL patients defined this way were superior to nodal NHL patients; however, the difference in OS was reversed after correction for differences in International Prognostic Index and malignancy grade.

Conclusion:

This study illustrates the selection bias that is introduced when a strict definition of primary extranodal NHL, that excludes cases with disseminated disease, is used. Patients with primary extranodal NHL were found to have a superior DFS, irrespective of which definition of primary extranodal NHL was used.

Key words: extranodal, nodal, non-Hodgkin’s lymphoma, prognosis


    Introduction
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
Most non-Hodgkin’s lymphomas (NHL) arise in lymph nodes or other lymphatic tissues such as the spleen, Waldeyer’s ring and thymus. Involvement of so-called extranodal organs is a common finding after staging investigation, however, and a substantial part of NHL even arises in these sites. The latter form is often referred to as primary extranodal NHL.

Although 25–40% of NHL patients present with a primary extranodal lymphoma [14], and numerous papers dealing with extranodal NHL originating in almost every organ in the body have been published, the literature on primary extranodal lymphomas as a group is limited. Moreover, the definition of primary extranodal lymphoma is a controversial issue, especially in patients where both nodal and extranodal sites are involved. In some series on primary extranodal NHL this problem is circumvented, and only patients with localised disease are included [2, 58]. As primary extranodal lymphomas have the potential to disseminate, these series thus present a picture of extranodal NHL that is incomplete. On the other hand, studies that use more liberal criteria for extranodal NHL [1, 4, 913], and include patients with disseminated disease, might erroneously include patients with disease that originated in lymph nodes or other nodal sites.

In the Comprehensive Cancer Centre West (CCCW) NHL registry, we were also confronted with the problems associated with the definition of primary extranodal NHL as outlined above, especially in patients with disseminated disease at both nodal and extranodal sites. For these patients, a third pattern of presentation was proposed: extensive disease [14, 15]. Data from our registry were used to define and explore the relevance of this subdivision of NHL in primary nodal, extranodal and extensive involvement. Its prognostic significance was tested, together with the two alternative subdivisions of nodal and extranodal NHL commonly used in the literature, by relating them to prognostic factors, response to therapy and survival.


    Patients and methods
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
Inclusion criteria and lymphoma classification
Included in the CCCW NHL registry were all newly diagnosed NHL patients living in the region covered by the CCCW (determined by postal code). Registration started on 1 June 1981 and ended on 31 December 1989. Once a diagnosis of NHL was made by the hospital pathologist, slides and frozen material were sent for review to a panel of four regional pathologists, experienced in hemato-pathology. Additional immunological and enzyme-histochemical techniques were performed to enable classification according to the Kiel classification, with some modifications, and the International Working Formulation. Only patients with a diagnosis of NHL confirmed by the pathology panel were included in the database, while patients with primary cutaneous lymphoma, multiple myeloma, acute lymphoblastic leukaemia and chronic lymphocytic leukaemia, and all cases diagnosed upon post-mortem examination were excluded from registration. After the inclusion of new patients had stopped, the Revised European–American Lymphoma Classification (REAL) for lymphomas was proposed by the International Lymphoma Study Group. As this new scheme for classification of lymphomas was rapidly accepted by lymphoma investigators throughout the world, it was decided to classify all cases according to the REAL classification, and to add this information to the dataset. Material of patients with diffuse large B-cell lymphoma (DLBL), mantle cell lymphoma, marginal zone extranodal NHL and Burkitt’s lymphomas was re-classified after revision. For other cases, the modified Kiel classification was directly converted into the REAL classification. In 140 cases, re-classification was not possible, therefore REAL classification was available for 1028 cases.

Staging
Clinical stage was defined according to the Ann Arbor classification [17]. Involvement of lymph nodes, spleen, thymus and Waldeyer’s ring were defined as nodal localisations, and the involvement of other organs was defined as extranodal. Patients were considered to be completely staged when adequate information was available on history, status of peripheral lymph nodes (physical examination), Waldeyer’s ring (examination by an ENT specialist), mediastinal lymph nodes (chest X-ray), abdominal lymph nodes, liver and spleen (abdominal CT scan, or lymphography supplemented with isotope liver and spleen scan in the early years of the registration), as well on peripheral blood and bone marrow (cytology and histology). The International Prognostic Index (IPI) was calculated according to the description by the International Non-Hodgkin’s Prognostic Factors Project for patients with all required parameters present.

Treatment and response classification
The NHL study group formulated therapeutic guidelines, but decisions on treatment were always at the discretion of the local physician. The general guidelines until 1989 were: radiotherapy for localised lymphomas; wait and see policy or, if symptomatic, chemotherapy: cyclophosphamide, vincristine and prednisone (CVP) chemotherapy or chlorambucil for stage III–IV low grade NHL; cyclophosphamide, hydroxydaunomycin, vincristine and prednisone (CHOP) chemotherapy for stage II–IV intermediate and high grade NHL, followed by radiotherapy for bulky tumours or persistent lesions. Lymphoblastic NHL was treated as acute lymphoblastic leukaemia.

No special guidelines were given for the treatment of primary extranodal lymphomas. Complete response (CR) to treatment was defined as the disappearance of all clinical, radiological or other evidence of disease. For the purpose of the present study, all patient records were checked for the results of staging investigations. Besides the presenting site (from which the diagnostic biopsy was taken), all other sites clinically involved in NHL were coded using the International Classification of Disease for Oncology (ICD-O, 1976). Lymphomas presenting in lymph node, Waldeyer’s ring, spleen or thymus, with no or only minor extranodal involvement after staging were categorised, according to the Ann Arbor classification, as primary nodal NHL, while those with presentation at other sites, with no or only regional lymph node involvement after staging were considered to be of extranodal origin. Patients with disseminated disease in both nodal and extranodal localisations, in which no decision could be made with respect to primary nodal or extranodal origin, were categorised as having extensive involvement [15, 16]. Our criteria for classification of NHL as primary nodal, extranodal or extensive are detailed in Table 1, together with the two common definitions of primary extranodal NHL mentioned in the Introduction. The characteristics of patients with extensive involvement, and the prognostic significance of the different definitions of primary extranodal NHL were explored. Chi-square tests were used to compare percentages in cross tabulations and the log-rank test was used to compare Kaplan–Meier survival curves. Overall survival (OS) was calculated from date of diagnosis until death (all causes) or last follow-up. Disease-free survival (DFS) was calculated for patients in CR only, from date of CR until recurrence (failure) or last follow-up (including death in CR). Cox regression analysis was used to correct for the confounding effect of differences in prognostic factors.


View this table:
[in this window]
[in a new window]
 
Table 1. Three alternative definitions of primary nodal and extranodal non-Hodgkin’s lymphoma (NHL)
 

    Results
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
Staging results of all 1168 patients included in the database are presented in Table 2. With respect to these data, it should be noted that 74% of patients in the CCCW database were completely staged, and in the other 26%, one or more of the investigations required was missing (usually bone-marrow biopsy).


View this table:
[in this window]
[in a new window]
 
Table 2. Involved sites for all patients (n = 1168)
 
Fifty-two per cent of patients presented with an involved lymph node, with the majority localised above the diaphragm. Although the spleen was involved in almost one-fifth of all patients, very few patients had an enlarged spleen as presenting symptom (<1%). NHL presenting in Waldeyer’s ring was found in 7% of patients. Using the Ann Arbor criteria for nodal and extranodal sites (and thus considering Waldeyer’s ring as a nodal site), 60% of patients presented with primary involvement of a nodal site. Patients who presented with extranodal NHL had localisations in a wide variety of organs; the digestive tract and bone marrow were the sites most frequently involved (Table 2). The involvement of nodal and extranodal sites revealed by staging investigations is also presented in Table 2, as is the absolute frequency of involvement for all nodal and extranodal sites for all patients. Using the first definition (Table 1), 635 patients (54%) had primary nodal NHL, 389 patients (34%) primary extranodal NHL, and 140 patients (12%) presented with extensive involvement. Using the second definition, and including patients with extensive involvement in the primary nodal category, the proportion of primary nodal NHL increased to 775 (66%), with the number of primary extranodal NHL patients constant at 389 (34%). After application of the third definition, the primary nodal and extranodal subgroups contained 933 patients (80%) and 231 patients (20%), respectively.

The characteristics of all patients are presented in Table 3, employing the subdivisions ‘primary nodal’, ‘primary extranodal’ and ‘extensive involvement’. The REAL classification of all lymphomas, also subdivided by primary nodal, primary extranodal or extensive involvement, is presented in Table 4. The relative frequency of follicular NHL in the nodal , extranodal and extensive group was 28%, 7% and 19%, respectively (nodal versus extensive, P = 0.04; extranodal versus extensive, P <0.001). For DLBL the relative frequencies were 34%, 50% and 40% in the nodal, extranodal and extensive group, respectively (nodal versus extensive, P = 0.2; extranodal versus extensive, P = 0.05).


View this table:
[in this window]
[in a new window]
 
Table 3. Patient characteristics (n = 1164)a according to primary nodal, extranodal or extensive status (definition 1)
 

View this table:
[in this window]
[in a new window]
 
Table 4. Histological classification (REAL) according to primary nodal, extranodal and extensive status (n = 1164)a
 
The use of the three definitions of primary nodal and extranodal NHL resulted in subgroups with different risk profiles according to the IPI (Table 5). As can be concluded from Table 5, as well as from Table 3, the patients categorised as extensive (definition 1) represent a poor prognostic subgroup, with high lactate dehydrogenase, disseminated disease and high stage. Consequently, when these patients were included in the primary nodal category (definition 2), IPI scores of this subgroup increased, indicating a poorer prognosis. This effect is more prominent when patients with disseminated disease presenting at an extranodal site are also included in the primary nodal category (definition 3), and consequently the prognostic profile of extranodal NHL patients is ameliorated when definition 3 is used instead of definition 2.


View this table:
[in this window]
[in a new window]
 
Table 5. Different risk profiles according to the International Prognostic Index (IPI) as a result of the use of the three definitions of primary nodal and extranodal non-Hodgkin’s lymphoma
 
Patients with primary extranodal NHL received regional treatment more frequently than patients with primary nodal NHL, regardless of which definition was used. This difference was most prominent using definition 3, clearly because only patients with localised disease presenting at an extranodal site qualify as primary extranodal NHL under this definition, and treatment for patients with localised disease usually consisted of regional radiotherapy at the time of the registration.

Using definition 1, CR rates of primary nodal and extranodal NHL patients did not differ, but were superior when compared with patients with extensive involvement (P <0.001; Table 6). The OS of primary nodal and extranodal patients defined according to definition 1 did not differ significantly (P = 0.06; hazard ratio 1.2); patients with extensive involvement had inferior OS (P <0.001) (Table 6; Figure 1). After correction for differences in IPI risk score and malignancy grade by Cox regression analysis, however, these differences disappeared. The DFS of primary extranodal NHL patients was superior to patients with primary nodal NHL and to those with extensive involvement (P <0.001) (Table 6; Figure 1); this difference remained after correction for IPI and malignancy grade.


View this table:
[in this window]
[in a new window]
 
Table 6. Treatment outcome according to definition of primary nodal or extranodal status
 


View larger version (24K):
[in this window]
[in a new window]
 
Figure 1. Definition 1: overall (A) and disease-free (B) survival.

 
Primary nodal and extranodal NHL patients defined according to definition 2 had comparable CR rates and OS (P = 0.9) (Table 6; Figure 2), also after correction for IPI and malignancy grade. DFS of primary extranodal NHL patients (definition 2) was superior to nodal NHL patients (P <0.001) (Table 6; Figure 2), also after correction for IPI and malignancy grade.



View larger version (20K):
[in this window]
[in a new window]
 
Figure 2. Definition 2: overall (A) and disease-free (B) survival.

 
As could be expected, patients with primary extranodal NHL defined according to definition 3 had a more favourable response to treatment and OS than those with primary nodal NHL (P <0.001) (Table 6; Figure 3). After correction for IPI and malignancy grade this difference was reversed: nodal NHL patients (definition 3) appeared to have a superior 5-year OS (P = 0.05; hazard ratio 1.45). DFS of extranodal NHL patients was superior when compared with nodal NHL, also after correction for differences in IPI and malignancy grade (P = 0.01; hazard ratio 0.61). When the analysis was restricted to patients with stage I and II disease (for both primary nodal and extranodal NHL patients), the same results were obtained.



View larger version (21K):
[in this window]
[in a new window]
 
Figure 3. Definition 3: overall (A) and disease-free (B) survival.

 

    Discussion
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
The definition of primary extranodal NHL is a controversial issue. In the Ann Arbor classification [17], no distinction is made between primary nodal and primary extranodal lymphomas, clearly because primary extranodal involvement is extremely rare in patients with Hodgkin’s disease, for whom this classification was originally designed. The limited validity of the Ann Arbor classification for NHL patients, as established by Rosenberg [18], and the observation by Musshoff and others that primary extranodal NHL constitutes a separate clinical entity with a relatively favourable prognosis was reason to propose a modification of the Ann Arbor classification to be used for NHL [19], in which primary nodal and extranodal NHL were classified separately. However, this proposal has never been widely accepted.

Following review of the literature it can be concluded that some authors use strict criteria to define primary extranodal NHL, whereas others use a more liberal definition. Strict criteria were proposed in 1961 by Dawson [20], reporting on a series of primary intestinal lymphomas. Patients were only included in this study if no palpable superficial lymphadenopathy was found at first examination, chest X-ray showed no obvious enlargement of mediastinal lymph nodes and the white blood cell counts were within normal limits. Furthermore, upon laparotomy the bowel lesion had to be shown to be dominant, and only involvement of lymph nodes in the immediate vicinity of the primary lesion was acceptable. Patients were excluded from analysis when distant abdominal lymph nodes, spleen or liver were involved. Several authors followed this definition (with the modification that no strict criteria are formulated on the white cell count), and included only patients with Ann Arbor stages I and II in their reports on primary extranodal NHL [2, 68].

Other investigators used more liberal criteria in their reports on primary extranodal lymphomas. Lewin et al. [21], describing a series of gastrointestinal tract lymphomas treated at Stanford University Medical Centre, included gastrointestinal lymphoma patients who presented with gastrointestinal symptoms, and also those who appeared to have disseminated disease upon staging. In an earlier report on extranodal NHL in the CCCW NHL registration [4], Otter et al. extrapolated this definition to all extranodal lymphomas. Others also used it, with the extra requirement that after routine staging procedures the extranodal component is clinically dominant [1, 10, 11].

In the present study, we used the CCCW population-based NHL registry to explore the effect of different definitions of primary extranodal NHL on the percentage of patients considered to have presented with primary extranodal disease, their treatment outcome and survival. At first, a subdivision in three presentation patterns was tested: patients with primary nodal, primary extranodal and extensive involvement were distinguished. As could be expected, the subgroup of patients with extensive involvement comprised only stage IV patients, with high IPI scores, and inferior treatment outcome when compared with primary nodal or extranodal NHL patients. When analysed within the IPI risk groups, however, the differences between patients with extensive involvement and those with primary nodal or extranodal NHL disappeared. This observation illustrates that separate classification of patients with extensive involvement does not provide prognostic information other than present in the IPI. Although misclassification of individual cases cannot be excluded, it seemed logical to include patients with extensive involvement in the primary nodal subgroup (according to definition 2). This choice is supported by the similarity of relative frequencies of follicular and diffuse large cell lymphomas in the primary nodal and extensive patient groups (Table 4). Using the second definition we found one-third of patients in the CCCW NHL registry presenting with primary extranodal NHL, a percentage that corresponds rather well with two other population-based studies in the literature. In a study from Denmark [1], where primary extranodal NHL was defined in a similar way, d’Amore et al. found 37% of patients presenting with extranodal disease. The National Cancer Data Base report on NHL [13], with data on >90 000 NHL patients retrieved from >1400 hospitals in the United States, used the site of NHL presentation as the only criterion for primary nodal or extranodal disease, and found 28% of patients presenting with a primary extranodal lymphoma. Three older, often-cited series on primary extranodal NHL, only including patients with localised disease (definition 3 in our study), found that 25% [2], 20% [7] and 10% [8] of patients presented with primary extranodal NHL, respectively. These percentages correlate rather well with the percentage of primary extranodal NHL we found in the CCCW NHL database, when patients with disseminated disease were excluded from the primary extranodal NHL group (definition 3), but probably underestimate the relative frequency of primary extranodal NHL cases in the general population. The favourable response rate and OS found for extranodal NHL patients defined in this strict way is, as we showed, the result of selection. With inclusion of disseminated cases in the extranodal group (definition 2), no differences in response rate and OS were found between primary nodal and extranodal patients. This is in accordance with the population-based study from the United States [13] mentioned previously, but contrary to the findings in the Danish population, where primary extranodal patients had slightly inferior OS when compared with primary nodal NHL patients [1].

An interesting finding in our study was the superior DFS found for patients with primary extranodal NHL that appeared independent of both the definition of primary extranodal NHL and differences in prognostic factors. It is possible that differences in biological behaviour of NHL originating in nodal and certain extranodal sites account for this observed difference in DFS. This possibility is strongly suggested by Gospodarowicz et al. in a study where patterns of disease in patients with localised nodal and extranodal lymphomas who only received regional radiotherapy were compared [5]. Patients with lymphomas originating in the gut-associated lymphatic tissue (GALT) were found to have a significantly lower distant relapse rate than patients with nodal lymphomas and patients with extranodal lymphomas from other sites than the GALT.

In conclusion, we explored the effect of different definitions of primary extranodal NHL in a population-based cohort of NHL patients. Our analysis showed that in ~10% of patients, a distinction between nodal or extranodal origin could not be made; a separate analysis of these patients suggested that they are probably best classified as cases with primary nodal NHL. This study also illustrates the selection bias that is introduced when a strict definition of primary extranodal NHL, which excludes cases with disseminated disease from further analysis, is used. We therefore propose to use a liberal definition of primary extranodal NHL that includes all patients who present with NHL that apparently originated at an extranodal site, even in the presence of disseminated disease, as long as the extranodal component is clinically dominant.


    Footnotes
 
+ Correspondence to: Dr A. D. G. Krol, Department of Clinical Oncology, K1-P, Leiden University Medical Centre, PO Box 9600, 2300 RC Leiden, The Netherlands. Tel: +31-71-5263027; Fax: +31-71-5266760; E-mail: a.d.g.krol{at}lumc.nl Back


    References
 Top
 Abstract
 Introduction
 Patients and methods
 Results
 Discussion
 References
 
1. d’Amore F, Christensen BE, Brincker H et al. Clinicopathological features and prognostic factors in extranodal non-Hodgkin lymphomas. Danish LYFO Study Group. Eur J Cancer 1991; 27: 1201–1208.[ISI][Medline]

2. Freeman C, Berg JW, Cutler SJ. Occurrence and prognosis of extranodal lymphomas. Cancer 1972; 29: 252–260.[ISI][Medline]

3. Newton R, Ferlay J, Beral V et al. The epidemiology of non-Hodgkin’s lymphoma: comparison of nodal and extra-nodal sites. Int J Cancer 1997; 72: 923–930.[CrossRef][ISI][Medline]

4. Otter R, Gerrits WB, Sandt MM et al. Primary extranodal and nodal non-Hodgkin’s lymphoma. A survey of a population-based registry. Eur J Cancer Clin Oncol 1989; 25: 1203–1210.[CrossRef][ISI][Medline]

5. Gospodarowicz MK, Sutcliffe SB, Brown TC et al. Patterns of disease in localized extranodal lymphomas. J Clin Oncol 1987; 5: 875–880.[Abstract]

6. Gospodarowicz MK, Sutcliffe SB. The Extranodal Lymphomas. Semin Radiat Oncol 1995; 5: 281–300.[Medline]

7. Paryani S, Hoppe RT, Burke JS et al. Extralymphatic involvement in diffuse non-Hodgkin’s lymphoma. J Clin Oncol 1983; 1: 682–688.[Abstract]

8. Rudders RA, Ross ME, DeLellis RA. Primary extranodal lymphoma: response to treatment and factors influencing prognosis. Cancer 1978; 42: 406–416.[ISI][Medline]

9. Arican A, Dincol D, Akbulut H et al. Clinicopathologic features and prognostic factors of primary extranodal non-Hodgkin’s lymphomas in Turkey. Am J Clin Oncol 1999; 22: 587–592.[CrossRef][ISI][Medline]

10. Economopoulos T, Asprou N, Stathakis N et al. Primary extranodal non-Hodgkin’s lymphoma in adults: clinicopathological and survival characteristics. Leuk Lymphoma 1996; 21: 131–136.[ISI][Medline]

11. Zucca E, Roggero E, Bertoni F et al. Primary extranodal non-Hodgkin’s lymphomas. Part 1: Gastrointestinal, cutaneous and genitourinary lymphomas. Ann Oncol 1997; 8: 727–737.[CrossRef][ISI][Medline]

12. Zucca E, Roggero E, Bertoni F et al. Primary extranodal non-Hodgkin’s lymphomas. Part 2: Head and neck, central nervous system and other less common sites. Ann Oncol 1999; 10: 1023–1033.[CrossRef][ISI][Medline]

13. Glass AG, Karnell LH, Menck HR. The National Cancer Data Base report on non-Hodgkin’s lymphoma. Cancer 1997; 80: 2311–2320.[CrossRef][ISI][Medline]

14. Kramer MH, Hermans J, Parker J et al. Clinical significance of bcl2 and p53 protein expression in diffuse large B-cell lymphoma: a population-based study. J Clin Oncol 1996; 14: 2131–2138.[Abstract]

15. Kramer MH, Hermans J, Wijburg E et al. Clinical relevance of BCL2, BCL6, and MYC rearrangements in diffuse large B-cell lymphoma. Blood 1998; 92: 3152–3162.[Abstract/Free Full Text]

16. The Non-Hodgkin’s Lymphoma Classification Project. A clinical evaluation of the International Lymphoma Study Group classification of non-Hodgkin’s lymphoma. Blood 1997; 89: 3909–3918.[Abstract/Free Full Text]

17. Carbone PP, Kaplan HS, Musshoff K et al. Report of the Committee on Hodgkin’s Disease Staging Classification. Cancer Res 1971; 31: 1860–1861.[ISI][Medline]

18. Rosenberg SA. Validity of the Ann Arbor staging classification for the non-Hodgkin’s lymphomas. Cancer Treat Rep 1977; 61: 1023–1027.[ISI][Medline]

19. Musshoff K. Classification of the clinical stages of non-Hodgkin’s lymphoma. Strahlentherapie 1977; 153: 218–221.[ISI][Medline]

20. Dawson IP, Cornes JS, Morson BC. Primary malignant lymphoid tumours of the intestinal tract. Report of 37 cases with a study of factors influencing prognosis. Br J Surg 1961; 49: 80–89.[ISI]

21. Lewin KJ, Ranchod M, Dorfman RF. Lymphomas of the gastrointestinal tract: a study of 117 cases presenting with gastrointestinal disease. Cancer 1978; 42: 693–707.[ISI][Medline]